Ships

Bao-Lin Zhang · Qing-Long Han
Xian-Ming Zhang · Gong-You Tang
Active Control
of Offshore
Steel Jacket
Platforms
Active Control of Offshore Steel Jacket Platforms
Bao-Lin Zhang • Qing-Long Han
Xian-Ming Zhang • Gong-You Tang
Active Control of Offshore

Steel Jacket Platforms
123
Bao-Lin Zhang Qing-Long Han
China Jiliang University Swinburne University of Technology
Hangzhou, Zhejiang, China Melbourne, VIC, Australia
Xian-Ming Zhang Gong-You Tang

Swinburne University of Technology Ocean University of China
Melbourne, VIC, Australia Qingdao, Shandong, China
ISBN 978-981-13-2985-2 ISBN 978-981-13-2986-9 (eBook)

https://doi.org/10.1007/978-981-13-2986-9
Library of Congress Control Number: 2018961387
© Springer Nature Singapore Pte Ltd. 2019

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
Offshore platforms are widely used to explore, drill, produce, store, and transport
ocean resources and are usually subjected to environmental loading, such as waves,
winds, ice, and currents, which may lead to failure of deck facilities, fatigue failure
of platforms, inefficiency of operation, and even discomfort of crews. In order to
ensure reliability and safety of offshore platforms, it is of great significance to
explore a proper way of suppressing vibration of offshore platforms. There are
mainly three types of control schemes, i.e., passive control, semi-active control, and
active control schemes, to deal with the vibration of offshore platforms. This book
provides a brief overview of these schemes and mainly presents recent advances of
active control schemes with optimal tracking control, sliding model control, delayed
feedback control, and network-based control.
Structure and readership This book consists of nine parts. An overview of
vibration control of offshore platforms is provided in Chap. 1, where passive control
schemes and several semi-active control schemes are briefly summarized; some
classical active control approaches, such as optimal control, robust control, and
intelligent control, are briefly reviewed; and recent advances of active control
schemes with optimal tracking control, integral sliding mode control, delayed
feedback control, and network-based control are deeply analyzed.
Dynamic models of offshore platforms Chapter 2 presents two dynamic models
of the offshore platforms and several mathematical lemmas used in this book. In
the first dynamic model, only the first dominant vibration mode of an offshore
steel jacket platform with an active mass damper (AMD) mechanism is taken into
account. The model is mainly utilized to design active controllers to attenuate
wave-induced vibration of the offshore platform. In the second dynamic model, the
first two dominant vibration modes of an offshore steel jacket platform subject to
an active tuned mass damper (TMD) mechanism are considered. By considering
parametric perturbations of the system and external disturbance, several uncertain
nonlinear models for the offshore platform are developed. Such models are utilized
to design active controllers to reduce vibration amplitudes of the offshore platform
subject to self-excited hydrodynamic forces and/or external disturbance.
v
vi Preface
Optimal tracking control Chapter 3 presents an optimal tracking control scheme

to suppress wave-induced vibrations of an offshore platform subject to an AMD
mechanism. Based on a linear exogenous system model of the external wave
force on the offshore platform, an optimal tracking controller with feedforward
compensation is proposed. The existence and uniqueness of the optimal tracking
controller are discussed. The computation process of controller gain is provided,
and the effectiveness of the controller is demonstrated via simulation results.
Integral sliding mode H∞ control In Chap. 4, an integral sliding mode H∞ control
scheme for an offshore steel jacket platform subject to nonlinear self-excited wave
force and external disturbance is developed. In the case where the dynamic model
of the offshore platform is subject to parameter perturbations, a robust sliding mode
H∞ control scheme is proposed. The effectiveness and advantages of the integral
sliding mode H∞ control scheme are investigated.
Delayed integral sliding mode control In Chap. 5, by intentionally introducing a
proper time-delay into the control channel, a sliding mode control scheme utilizing
mixed current and delayed states is proposed. Based on simulation results, the
effectiveness and superiority of the proposed controller are analyzed, and the
positive effects of intensionally introduced time-delays on the sliding mode control
for the offshore platform are investigated.
Delayed robust non-fragile H∞ control This chapter provides a delayed non-fragile
H∞ control scheme for an offshore steel jacket platform subject to self-excited
nonlinear hydrodynamic force and external disturbance. A delayed robust non-
fragile H∞ controller is designed to reduce the vibration amplitudes of the offshore
platform. The positive effects of time-delays on the non-fragile H∞ control for
the offshore platform are investigated. Simulation results show that the proposed
delayed non-fragile H∞ controller is effective to attenuate vibration of the offshore
platform; the control force required by the delayed non-fragile H∞ controller is
smaller than the one by the delay-free non-fragile H∞ controller; and the time-
delays can be used to improve the control performance of the offshore platform.
Delayed dynamic output feedback control Chapter 7 discusses the design process
and computation algorithm of a delayed dynamic output feedback controller to
reduce wave-induced vibration of an offshore platform. A conventional dynamic
output feedback controller is designed first to reduce the internal oscillations of the
offshore platform. It is observed that the designed controller is of a larger gain in the
sense of Euclidean norm, which demands a larger control force. Then, a time-delay
is introduced intentionally to design a new dynamic output feedback controller such
that the controller is of a small gain in the sense of Euclidean norm; and the internal
oscillations of the offshore platform can be dramatically reduced.
Network-based modeling and active control Offshore platforms are generally
located far away from land and always affected by complicated and harsh ocean
environmental loads. Thus, network-based control paves an effective way to lower
control costs and simplify installation and maintenance of an offshore platform,
Preface vii
while the safety of staff on the platform can be ensured. Chapters 8 and 9 focus on
dealing with network-based modeling and network-based control issues of offshore
platforms. In Chap. 8, for an offshore steel jacket platform with an active TMD
mechanism, a network-based state feedback control scheme is developed. Under
this scheme, the corresponding closed-loop system is modeled as a system with
an artificial interval time-varying delay, and a sufficient condition on the existence
of the network-based controller is obtained. The effects of network-induced time-
delays on the performance of the offshore platform are investigated.
Event-triggered H∞ reliable control in network environments In Chap. 9, a
network-based model of the offshore platform subject to external wave force
and actuator faults is presented, and an event-triggered H∞ reliable controller is
designed such that during the control implementation, only requisite sampled-data
are transmitted over networks. It is observed that the networked controllers are
capable of guaranteeing the stability of the offshore platforms and reducing the
required control cost. Moreover, the proposed network-based controllers are better
than some existing ones without network setting.
Acknowledgment We would like to acknowledge the collaborations with Profes-
sor Xinghuo Yu on the work of sliding mode control reported in this monograph. The
supports from the Key Project of Natural Science Foundation of Zhejiang Province
of China under Grants Z19F030002, the National Natural Science Foundation of
China under Grants 61773356, 61379029, and 61673357, the State Foundation for
Studying Abroad under Grant 201308330318, the Scientific Research Foundation
for the Returned Overseas Chinese Scholars, Ministry of Education of China under
Grant 2012-1707, the Natural Science Foundation of Zhejiang Province under
Grant Y1110036, the Academic Climbing Foundation of Youth Discipline Leaders
of Universities in Zhejiang Province under Grant PD2013190, the Australian
Research Council Discovery Projects under Grants DP1096780, DP0986376, and
DP160103567, and the Griffith University 2016 New Researcher Grant Scheme
under Grant 219128 are gratefully acknowledged. Finally, the close cooperation
with Springer as publisher and particularly with Dr. Jasmine Dou as Associate
Editor is gratefully acknowledged.
Hangzhou, China Bao-Lin Zhang

Melbourne, VIC, Australia Qing-Long Han
Melbourne, VIC, Australia Xian-Ming Zhang
Qingdao, China Gong-You Tang
December 2018
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Passive Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Hysteretic and/or Viscoelastic Mechanisms . . . . . . . . . . . . . . . . . . . 2
1.1.2 Damping Isolation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 Dynamic Vibration Absorbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Semi-active Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Active Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Optimal Control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.2 Robust Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.3 Sliding Mode Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.4 Delayed Feedback Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.5 Network-Based Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Book Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 Dynamic Models of Offshore Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 Model of an Offshore Platform with AMD Mechanisms . . . . . . . . . . . . . 17
2.2 Model of an Offshore Platform with Active TMD Mechanisms. . . . . . 22
2.3 Some Related Mathematical Lemmas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3 Optimal Tracking Control with Feedforward Compensation. . . . . . . . . . . 33
3.1 System and Problem Descriptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Design of Optimal Tracking Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Simulation Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.1 Performance of System with Optimal Tracking Controller . . . 39
3.3.2 Comparison of Optimal Controller and Tracking
Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
ix
x Contents
4 Integral Sliding Mode H∞ Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

4.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 Sliding Surface Design and Stability Analysis of Sliding Motion . . . . 50
4.3 Design of the Sliding Mode H∞ Control Law . . . . . . . . . . . . . . . . . . . . . . . . 52
4.4 Design of the Robust Sliding Mode H∞ Control Law . . . . . . . . . . . . . . . . 54
4.5 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.5.1 System Parameters of an Offshore Platform. . . . . . . . . . . . . . . . . . . 56
4.5.2 Performance of the Nominal System . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.5.3 Performance of the Uncertain System . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.7 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5 Delayed Integral Sliding Mode Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.1 Design of Delayed Robust Sliding Mode Controllers . . . . . . . . . . . . . . . . . 71
5.1.1 Integral Sliding Surface Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.1.2 Design of a Robust Delayed Sliding Mode Controller . . . . . . . . 75
5.1.3 Design of a Delayed Sliding Mode Controller . . . . . . . . . . . . . . . . 76
5.2 A Computational Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.3.1 Simulation Results for the Nominal System. . . . . . . . . . . . . . . . . . . 79
5.3.2 Simulation Results for the Uncertain System . . . . . . . . . . . . . . . . . 83
5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6 Delayed Robust Non-fragile H∞ Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.2 Design of a Delayed Robust Non-fragile H∞ Controller . . . . . . . . . . . . . 92
6.3 Simulation Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.1 Performance of the Nominal System . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.2 Performance of the Uncertain System . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.3.3 Comparison of Several Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7 Delayed Dynamic Output Feedback Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.1 Dynamic Output Feedback Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.2 Design of a Delayed Dynamic Output Feedback Controller . . . . . . . . . . 115
7.2.1 Delayed Dynamic Output Feedback Controller Design . . . . . . 116
7.2.2 A Computational Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.2.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.3 Comparison Between Different Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
7.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Contents xi
8 Network-Based Modeling and Active Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

8.2 Stability Analysis and Network-Based Controller Design . . . . . . . . . . . . 134
8.3.1 Performance of System with Network-Based Controllers . . . . 140
8.3.2 Comparison of Controllers With and Without Network
Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8.3.3 Effect Analysis of Network-Induced Delays . . . . . . . . . . . . . . . . . . 150
8.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
8.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
9 Event-Triggered H∞ Reliable Control in Network Environments . . . . 155
9.2 Design of an Event-Triggered H∞ Reliable Controller . . . . . . . . . . . . . . . 159
9.3.1 Event-Triggered H∞ Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
9.3.2 Event-Triggered H∞ Reliable Control: Constant Delays . . . . . 174
9.3.3 Event-Triggered H∞ Reliable Control: Time-Varying
Delays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
9.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
9.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Acronyms
A System matrix
A−1 Inverse of matrix A
AT Transpose of matrix A
A≥0 Symmetric positive semi-definite
A>0 Symmetric positive definite
A≤0 Symmetric negative semi-definite
A<0 Symmetric negative definite
det (A) Determinant of matrix A
diag(X1 , X2 , · · · , Xm ) Diagonal matrix with Xi as its ith diagonal element
I Identity matrix of appropriate dimensions
lim Limit
LMI Linear matrix inequality
N Positive integers
R Field of real numbers
Rn n-dimensional real Euclidean space
Rn×m Space of n × m real matrices
sgn(x) The sign of x
tr(A) Trace of matrix A
h The sampling interval of sensor
0n×m Zero matrix of dimension n × m
λ(A) Eigenvalue of matrix A
x(k) The state variable vector at time kT
|x| Absolute value (or modulus) of x
x Euclidean norm
P Induced norm supx=1 P x
∀ For all
∈ Belong to
→ Tend to, or mapping to (case sensitive)
xiii
xiv Acronyms
⊗
Matrix Kronecker product
sum
sup Supremum
inf Infimum
∗ Entries implied by symmetry
Chapter 1
Introduction
Offshore platforms are extensively used to explore, drill, produce, storage, and
transport ocean oil and/or gas resources in different depths. There are several
types of offshore platforms, such as self-elevating platforms, gravity platforms,
steel jacket platforms, tension leg platforms (TLPs), articulated leg platforms,
guyed tower platforms, spar platforms, floating production systems, and very large
floating structures. These platforms can be divided into fixed-bottom platforms
and buoyant platforms, which have their own particular purposes and different
configurations. To meet an increasing demand for marine sources of energy and
minerals, in the past several decades, a lot of research effort has been made on
offshore platforms. The related investigations are focused mainly on structure design
and monitoring, damage detection, fatigue analysis and reliability assessment,
mathematical modeling, and analysis of structures. Specifically, offshore platforms,
which are located in a very tough ocean environment over a long period of time, are
inevitably affected by environmental loading, such as waves, winds, ice, currents,
flow, and earthquakes [1, 2]. The environmental loading may lead to excessive
vibration of offshore platforms, thereby causing failure of deck facilities, fatigue
failure of structures, inefficiency of operation, and even discomfort of crews. Note
that reduction of vibration amplitude of an offshore platform by 15% can extend
service life over two times and can result in decreasing expenditure on maintenance
and inspection of structures [3]. Therefore, it is of great significance to explore
proper ways to reduce different types of vibrations of offshore platforms, and
comprehensive surveys of vibration control for offshore structures are provided by
Kandasamy et al. [4] and Zhang et al. [5].
Notice that a direct and simple way to mitigate vibration of offshore platforms
is to increase the stiffness of the platforms. As a result, natural frequencies can
be shifted away from resonating frequencies [6]. However, such schemes generally
require extra construction material, which unavoidably leads to increasing costs.
Thus, an alternative way is to choose a proper structural control method to reduce
structural vibration to an acceptable level [7, 8]. In the past several decades,
© Springer Nature Singapore Pte Ltd. 2019 1

B.-L. Zhang et al., Active Control of Offshore Steel Jacket Platforms,
https://doi.org/10.1007/978-981-13-2986-9_1
2 1 Introduction
structural control schemes, such as passive control schemes [9], semi-active control
schemes [10], and active control schemes [11], are widely utilized to reduce
vibration of offshore platforms.
This chapter provides an overview of recent advances in vibration control of off-
shore platforms. Firstly, some passive control schemes including hysteretic mech-
anisms, viscoelastic mechanisms, damping isolation mechanisms, and dynamic
vibration absorbers utilized in vibration control are briefly outlined. Secondly, semi-
active control schemes are briefly surveyed. Thirdly, active control schemes are
reviewed in detail. Several active control schemes with optimal control, robust
control, intelligent control, sliding mode control, and sampled-data control are
presented. In particular, effectiveness and superiority of delayed feedback control
and network-based control, which are most recently developed active control
schemes, are deeply discussed. By purposely introducing time-delays into control
channel of offshore platforms, delayed feedback control schemes focus on the
controller design and investigate the effects of time-delays on active control for
the platforms. Network-based control schemes are concerned with network-based
modeling and controller design of offshore platforms in network settings.
1.1 Passive Control
Passive control does not require any other external energy, and control force is
generally derived from deformation of devices themselves, such as different pas-
sive energy dissipation mechanisms, damping isolation mechanisms, and dynamic
vibration absorbers [6]. Due to easiness and lower cost to implement and retrofit,
effectiveness to mitigate vibration and to maintain stability, and reliability of
offshore platforms, passive control has been implemented for offshore platforms
extensively and successfully in the last two decades. This section gives a brief review
of passive control for offshore platforms.
1.1.1 Hysteretic and/or Viscoelastic Mechanisms
Hysteretic devices include metallic dampers and friction dampers, which dissipate
energy with no significant rate dependence. Viscoelastic devices are dependent on
frequency. The main types of viscoelastic devices are viscoelastic solid dampers
and viscoelastic fluid dampers. To mitigate vibration of offshore platforms and to
avoid damage of the offshore platforms, hysteretic and/or viscoelastic systems are
applied to the offshore platforms extensively. To mention a few, seismic response
of an offshore platform with shape memory alloy dampers is investigated in [12],
where influence of the number, location, and property of the shape memory alloy
dampers on vibration amplitudes of the offshore platform is discussed. It should be
1.1 Passive Control 3
specified that as an intelligent material, the shape memory alloy is of advantages

of preferable endurance and erosion resistance, good fatigue resistance, long-term
reliability, large elastic deformation, steady mechanical properties in temperature,
and frequency. Therefore, using the shape memory alloy as a damper is suitable
for offshore platforms. In [9], viscoelastic dampers are adopted to reduce wave-
induced vibration of an offshore steel jacket. It is observed that vibration amplitudes
of the platform can be reduced dramatically. By modeling an offshore platform as a
multi-degrees-of-freedom system subject to wave forces, vibrations of the offshore
platform with friction, viscoelastic, and viscous dampers are investigated, respec-
tively [13]. It is found that added viscoelastic dampers increase viscous damping
as well as lateral stiffness, and viscoelastic dampers are more effective to reduce
wave-induced vibration of the offshore platform than viscous and friction dampers.
By using a stochastic linearization technique, nonlinear behavior of friction dampers
installed in an offshore platform is discussed in [14]. The results show that optimally
regulating friction dampers leads to considerable reduction of topside displacement
and velocity variances for deep water platforms, while efficiency of such devices
decreases for shallow water platforms. Note that an energy dissipation rate in
friction dampers is displacement-dependent, and damping force is independent
of velocity response of the structure and frequency of excitations. Consequently,
friction dampers are specifically suitable for low-frequency excitations. In [15], by
using a hydrodynamic buoyant mass damper, which utilizes damper buoyancy and
inertia forces, and hydrodynamic damping effects, displacement responses of an
offshore platform subject to waves are investigated.
1.1.2 Damping Isolation Mechanisms
To dissipate unwanted vibration of offshore platforms, designing an isolation

mechanism placed between the bottom of deck and above the structure is also
one of simple and effective ways. For example, a damping isolation mechanism
composing of rubber bearings and viscous dampers is developed to suppress
vibration of offshore platforms subject to earthquake excitations and ice loads [3].
Both experimental and numerical studies are carried out to validate effectiveness in
suppressing vibration of the platforms. In [16], by using eight lead rubber bearings
as isolators, a lead rubber isolation mechanism is developed to mitigate earthquake-
induced vibration of a steel jacket offshore platform. Based on a platform in the
Persian Gulf, numerical studies are made to show effectiveness of the proposed
isolation mechanism. Note that the vibration acceleration of deck is an important
factor influencing human comfort and fatigue life of tubular joints [17]. In [18], to
reduce ice-induced vibration acceleration and avoid non-structure failure of jacket
platforms, an isolation cone mechanism composing of cones, springs, and dampers
is developed and installed between ice sheets and piles of jacket platforms.
4 1 Introduction
It is demonstrated from experimental and computational results that the proposed

damping isolation strategies are very effective to reduce vibration of offshore
platforms. However, due to the fact that damping isolation strategies are generally
utilized on newly built offshore platforms, to maintain and retrofit such mechanisms
are not always economical and convenient.
1.1.3 Dynamic Vibration Absorbers
Note that using damping dissipation devices to dissipate vibration energy generally
requires large relative deformation of the devices. However, relative deformation
of damping devices is not always large enough, which makes vibration reduction
unsatisfactory. In this case, dynamic vibration absorbers are feasible options. By
transferring some vibration energy to absorbers, an energy dissipation demand on
dominant vibration modes of offshore platforms is accomplished. The basic two
types of dynamic vibration absorbers are tuned mass dampers (TMDs) and tuned
liquid dampers (TLDs). In the past decades, TMD and TLD mechanisms have been
extensively implemented to mitigate vibration of offshore platforms.
1.1.3.1 TMD Mechanisms
A TMD device is composed of a mass, a spring, and a viscous damper mechanism.

It is generally installed in primary vibrating structures to suppress vibration of the
structures [19]. Due to remarkable advantages, such as effectiveness of refraining
vibration and easiness of implementation, TMD mechanisms are extensively used
for suppressing vibration of offshore platforms. In [11], a TMD is used to reduce
vibration of an offshore steel jacket platform subject to self-excited hydrodynamic
forces. In [20], a TMD device is installed inside columns of a TLP hull to attenuate
amplitudes of heave motion. In [21], based on a simplified single-degree-of-freedom
system of an offshore steel jacket platform with a TMD, energy dissipation and
transmission issues of the offshore platform are investigated, and an optimized
TMD design for the system is presented. In [22], a TMD is installed at the bottom
of deck plate of a multi-legged articulated tower to decrease its bending moment.
The feasibility of using a TMD device to mitigate ice-induced vibration of offshore
platforms in the Bohai Bay is studied experimentally [23]. The results show that
an optimally tuned TMD can attenuate ice-induced vibration of offshore platforms
favorably.
Note that a single TMD is generally effective to attenuate a single vibration mode
of structural systems. To guarantee that more than one distinct vibration modes can
be influenced, multiple TMDs are then introduced [24, 25]. In [26], the effectiveness
of multiple TMDs on suppressing wave-induced vibration of an offshore platform is
studied. The multiple TMDs are also used to mitigate the vibration of a TLP under
random wave forces. The vibration amplitudes of heave and pitch motion [27] as
1.1 Passive Control 5
well as surge motion [28] are investigated, respectively. It is shown that multiple
TMDs outperform a single TMD in terms of reducing oscillation amplitudes of
offshore platforms.
As aforesaid that TMD mechanisms have good control effect. However, such
methods generally add additional loads to offshore platforms. Therefore, applying
such schemes to deep-water platforms and flexible platforms is not always feasible.
To overcome disadvantage of TMDs where an additional mass is required to be the
mass body of a damper, an extended TMD and multiple extended TMD mechanisms
are proposed to reduce vibration of offshore platforms subject to earthquake and
wave loads, respectively [29, 30].
In [31], a TMD mechanism and a friction damper are applied to reduce wave-
induced vibration of an offshore platform, respectively. It is indicated that the
TMD and friction damper are both efficient in mitigating fatigue damage of the
offshore platform. It is also found that the former is more dependent on the dynamic
characteristics of the platform and the latter is more efficient for fixed steel jacket
platforms. In [32], by combining a TMD with a friction damper device, a hybrid
damping system is developed to control wave-induced and seismic vibrations and
fatigue damage of offshore platforms. In [33], based on the principle of TMD
mechanisms, an anti-vibration device is designed and installed on a similar model
of a jacket platform to control vertical and horizontal vibrations, respectively.
1.1.3.2 TLD Mechanisms
Due to effectiveness of vibration mitigation, low cost of manufacturing, convenience

of installation and maintenance, and availability of liquid, TLD devices are intro-
duced to control vibration of offshore platforms. In this control mechanism, liquid,
preferably water, is used as a damping mass, and liquid motion inside the damper
leads to oscillations in counter phase to structures and corresponding damping
effects [34], i.e., the system generally utilizes water sloshing force to attenuate
vibration of the structures. Specifically, such a system can be used as an additional
water reserve for building water supply and firefighting.
The earliest TLD device used in vibration control of offshore platforms can be
found in [35], where a liquid storage tank is placed on an offshore platform to reduce
wind-induced vibration. In [36], a liquid storage container is mounted on the top of
a fixed offshore platform and is used as a TLD device to suppress vibration induced
by earthquakes. Based on theoretical analysis, scale model test, and full-scale field
experiment, TLD containers are designed to reduce ice-induced vibration of a Bohai
production jacket platform, i.e., JZ20-2 MUQ platform [37]. In [38], a cylinder TLD
is applied to reduce vibration of a jacket platform subject to earthquakes. It is found
that the ratio of fundamental sloshing frequency of liquid to natural frequency of the
structure is a key factor to control seismic vibration. In [39–41], by using air springs
and water columns, tuned oscillators are presented to suppress vertical motion of
a deepwater TLP. The oscillators, which are mounted on the exterior of the TLP
columns, consist of several vertical caissons, which are open at the bottom of the sea
6 1 Introduction
and closed on top of a dual chamber. It is shown theoretically and experimentally

that the oscillators can obtain considerable reduction of vibration, while the mass of
platform does not increase.
By adopting shape of water tanks, the properties of TLDs can be changed
easily [34]. For instance, a traditional TLD device can be extended as a tuned
liquid column damper, which is a U-shape tube-like device and dissipates energy
by liquid flow between two liquid columns [42, 43]. The tuned liquid column
damper requires no mechanical components. Therefore, it is easy to retrofit the
damper without providing additional mass and to modify its natural frequency and
damping characteristics. Such a mechanism attracts extensive attention in vibration
control of structures [43]. In [44], active control of a tension-leg type of platform
incorporated with a tuned liquid column damper is studied both analytically and
experimentally. The effectiveness of reducing surge wave motion of the platform in
terms of vibration amplitudes and resonant frequencies is demonstrated.
In [45], a circular tuned liquid column damper is installed on an offshore platform
to mitigate structure vibration excited by ground motions. In [46], by replacing
an orifice in the tuned liquid column damper with a coated steel ball, which is
immersed inside a horizontal column of the damper, a tuned liquid column ball
damper is presented. Such a device is also used to mitigate wave-induced vibration
of an offshore platform [47]. It is shown through simulation results that the ball
damper performs better than the tuned liquid column damper. In [48], by replacing
an orifice tube with a smaller horizontal tube and combing a tuned liquid column
damper with buoyant members of a TLP, a water tuned liquid column damper
system is developed. Experimental results show that wave-induced vibration and
measured tension forces of TLP tethers can be reduced significantly. In addition,
structural safety can be ensured for the TLP with water tuned liquid column damper
mechanisms. In [49, 50], a tuned liquid column-gas damper presented by [51, 52] is
introduced to control vibration of offshore jacket platforms under seismic excitation.
To mention a few, tuned liquid column dampers adjust frequency by only tuning
length of the columns, while liquid column vibration absorbers adjust frequency by
tuning columns cross section. However, column-gas dampers adjust frequency by
tuning gas pressure in vertical columns as well as column length and cross section.
It is found that column-gas dampers can reduce root mean square (RMS) values
of vibration amplitudes significantly. To deal with restriction of a large horizontal
length using conventional tuned liquid column dampers, in [53], an S-shaped tuned
liquid column damper is proposed to suppress horizontal motion and vertical in-
plane rotation of a deep-sea floating platforms.
1.2 Semi-active Control
Note that semi-active control is a characteristic of controllable damping over contin-

uous range, fast response, and potentially low power requirements [4]. Combining
reliability of passive control and several advantages of active control, semi-active
1.3 Active Control 7
control is widely used in engineering vibration areas. Specifically, it requires less

energy to obtain better control effect [54]. Because of high reliability, inherent
stability, insensitivity to temperature, and small power requirement, magnetorheo-
logical (MR) dampers receive considerable attraction in vibration control of offshore
platforms. Generally, an MR damper is composed of a hydraulic cylinder, which
contains micro-sized particles suspended with a fluid. The damper characteristics of
the MR can be generated by adjusting the strength of magnetic field [55].
In [10], an MR damper is applied to suppress wave-induced vibration of an
offshore platform. Simulation results indicate that semi-active control with the MR
damper can reduce the maximum and the RMS values of vibration amplitude of the
platform effectively. To attenuate wave-induced vibration of an offshore platform
subject to hydrodynamic forces, semi-active control strategy using MR dampers
is verified for the offshore platform experimentally and numerically [56, 57]. In
[58, 59], a fuzzy MR controller is designed to control an offshore platform subject
to wave forces, and the effectiveness of the semi-active controller is investigated
numerically and experimentally. Located in Bohai Gulf of China, a JZ20-2NW
offshore platform is the first platform structure with an MR damper. By using
signals of main deck acceleration and isolation layer deformation, a Kalman filter is
designed to estimate system states of the platform first, and then ice- and seismic-
induced vibration responses of the offshore platform are investigated [60]. A linear
quadratic Gaussian method only using structural acceleration signal is developed
to reduce vibration of a steel jacket platform with an MR damper [61]. Simulation
results show that such a semi-active control strategy can attenuate vibration of the
offshore platform under random wave excitation. By combining feedforward neural
networks and fuzzy control methods, a neuro-fuzzy controller is proposed to control
wave-induced vibration of an offshore platform equipped with an MR damper [62],
where a neural network is used to approximate the nonlinear dynamic system and a
fuzzy logic controller is adopted to determine control force of the MR damper. It is
demonstrated that an intelligent semi-active controller is more effective than a linear
quadratic Gaussian optimal controller with an MR damper. In [63], a combination
of four MR dampers and six friction pendulum isolators on the joints of a cellar
deck is developed to control seismic vibration of a jacket offshore platform.
1.3 Active Control
In practical applications, a passive control device usually exhibits its limitations,

such as aging and endurance, long-term working reliability, and renovation or
substitution after strong shocks [12]. In addition, the control force is not adjustable
in real time [4]. Consequently, passive control should give way to active control,
which can achieve better control performance due to an introduction of external
control energy. Also, compared with passive control, active control has lower weight
and volume, wider bandwidth, and even potential to perform self-diagnosis [4].
Generally, an active control device consists of an actuator unit, a sensor unit, and
8 1 Introduction
a control unit. In the past several decades, active control has gained increasing
attention both in theory and in practice, and various active control mechanisms
are proposed for offshore platforms, to mention a few, active mass damper (AMD)
mechanisms, active tendon mechanisms [11, 64], active tuned mass damper (TMD)
mechanisms [65, 66], propeller thruster mechanisms [64, 67–69], and even their
combinations [70]. Based on these mechanisms, a number of efficient active
schemes are reported in the literature, which are given in detail as follows.
1.3.1 Optimal Control
As an advanced control technology, optimal control aims at optimizing a certain

objective function which is related to performance of a control system. In the
last two decades, optimal control is widely applied to active control of offshore
platforms. For example, in [71], a linear quadratic optimal control scheme is
used to suppress wave-induced vibration of an offshore steel jacket platform with
an active TMD mechanism, where a hydraulic servomechanism is designed to
regulate motion of the damper by designing an optimal controller. By taking into
consideration the first and second vibration modes of an offshore platform with an
active TMD mechanism, a two-loop feedback design scheme with linear quadratic
control theory is proposed for an offshore platform subject to hydrodynamic forces
[72]. In this control framework, an inner loop is designed to regulate the linear
part of the platform dynamics, while an outer loop is developed to accommodate
nonlinearities and maintain stability of the overall system. In [73], a classical linear
optimal controller in the frequency domain is presented. It is observed that an active
TMD is capable of reducing vibration of an offshore platform significantly. By
treating wave forces as output signals of a linear filter, a nonlinear stochastic optimal
control strategy for a wave-excited jacket platform is developed [74].
For an articulated leg platform under random sea state, a linear quadratic optimal
controller is designed to reduce vibration of an offshore platform [75]. Compared
with existing optimal control methods, an involved performance function depends
on displacement, velocity, as well as acceleration of the platform. Moreover,
by using an iterative frequency domain technique, system nonlinearities are also
considered. In [76], an optimal control method is applied to a TLP to control coupled
dynamic responses of the low-frequency hull motion and the bending strain of the
system tendons. The effectiveness of the control scheme is verified experimentally
based on a 1/100 scale model of a Japan Ocean Industries Association TLP. In
[77], an optimal controller is designed to control nonlinearly coupled responses of
a TLP subject to random sea wave forces and wind loads. By treating a TLP as a
tridimensional rigid body with six degrees of freedom and regarding the tendons
as elastic springs, an optimal controller is proposed to reduce heave displacement
amplitudes [78].
For jacket-type offshore platforms subject to wave forces, dynamic responses are
primarily dependent on the dominant frequency of exciting wave forces and the first
natural frequency of offshore platforms [11]. Based on a single-degree-of-freedom

model with an AMD mechanism [79], where only the first vibration mode of an
offshore steel jacket platform is considered, a feedforward and feedback optimal
control (FFOC) scheme is presented to attenuate vibration of the platform subject
to regular wave forces [80]. Further, by modeling irregular wave forces acting on
an offshore platform as output signals of a linear exogenous system, a discrete-
time FFOC scheme is developed [81]. Taking actuator time-delays into account,
a discrete FFOC controller with memory is designed for an offshore platform [82],
where the platform system with control delays is transformed into a non-delay linear
system. It is found from [80, 81] and [82] that the designed FFOCs are effective to
reduce wave-induced vibration of the platform. In [83] and [84], optimal tracking
control strategies are introduced; both continuous feedforward and feedback optimal
tracking controllers and discrete-time feedforward and feedback optimal tracking
controllers are developed, respectively. In [85], by modeling dynamic characteristics
of actuator faults as an output of a linear exogenous system, an observer-based
optimal fault-tolerant controller for an offshore platform is designed such that
reliability of the offshore platform with actuator faults can be ensured.
From the above analysis, it is clear that optimal control paves an effective way to
attenuate vibration of offshore platforms to a safe level. However, some existing
results on optimal control usually require that models of offshore platforms are
exactly known, which is not the situation in reality due to the fact that uncertain
external loading is frequently imposed on the platforms. It is still challenging to
extend those results to uncertain and general dynamic models of offshore platforms,
which is a significant research topic in the future.
1.3.2 Robust Control
Robust control is to design proper state/output feedback controllers against uncer-

tainties and external disturbance such that the worst-case value of some cost criteria
related to offshore platforms can be minimized. In [67] and [70], optimal frequency
domain approaches are developed to suppress wave-excited vibration of an offshore
platform with AMD mechanisms, where the H2 norm of transfer function from wave
forces to regulated outputs is minimized. It is shown that control devices are useful
in reducing amplitudes of displacement response of the platforms. For a simplified
offshore steel jacket platform with AMD mechanisms, H2 control schemes are
proposed to suppress wave-induced vibration of the system [79, 86, 87]. In [88],
a robust mixed H2 /H∞ control method is presented to attenuate wave-excited
vibration of the platform. A pure delayed state feedback H∞ control scheme is
introduced to control an offshore platform [89], where the required control force and
the vibration amplitudes of the offshore platform with a traditional H∞ controller, a
discrete FFOC [81], as well as a delayed H∞ controller are investigated. In [90],
a robust sliding mode H∞ controller is designed for an offshore platform with
an active TMD mechanism to reduce vibration excited by hydrodynamic forces
10 1 Introduction
and external disturbance. Recently, by considering controller perturbations caused

by physical limitations, component aging, or failure, a delayed non-fragile H∞
controller is developed to improve the performance of an offshore platform [91].
Notice that offshore platforms are a class of highly coupled nonlinear uncertain
systems subject to external disturbance. It is of importance to guarantee the stability
and reliability of offshore platforms from the practical point of view. A remarkable
advantage of robust control is robustness against discrepancies between a developed
model and a real system, parametric perturbations, and external disturbance.
Based on an ideal and simplified model, some nice results are reported in the
literature using robust H2 and/or H∞ control schemes. However, most of them
are mainly based on steel jacket platforms subject to external wave forces and/or
nonlinear hydrodynamic forces, while few of them are based on TLPs, self-elevating
platforms, gravity platforms, spar platforms, or very large floating structures.
In recent years, applications of soft computing, such as artificial neural networks
and fuzzy logics, attract increasing attention in the field of structural vibration
control. Due to several advantages including the ability to handle nonlinear behav-
iors, the inherent robustness, fault tolerance, and generalized capability, neural
networks and fuzzy logics are introduced to active control for offshore platforms.
In [92], back-propagation neural network-based active control is proposed to
suppress vibration of an offshore platform subject to random waves. In such a
control framework, a neural network is trained off-line with data from numerical
results, which are based on a classical linear quadratic optimal control method. A
robust control method using a modified probabilistic neural network is applied to
attenuate vibration of a fixed offshore platform under random waves [93]. Recently,
neural network-based controllers are designed to moderate vibration of an offshore
platform under earthquakes [94] and wind-induced random ocean waves [95],
respectively. In [96], based on a combination of gray prediction, support vector
machines, a dynamic stiffness matrix method, and an adoptive inverse control
method, an adaptive inverse controller is developed to decrease vibration of offshore
platforms subject to wave and wind loads.
It is observed that network control and fuzzy control strategies are practical and
efficient to deal with an active control problem of offshore platforms. However,
except for the aforesaid intelligent control schemes, few results are reported for
offshore platforms. Thus, further research exploring new intelligent control schemes
for different types of offshore platforms subject to complicated ocean environment
loading is expected in the near future.
1.3.3 Sliding Mode Control
It is well known that offshore platforms involve some uncertainties such as unknown
system parameters and structure flexibility. Most importantly, offshore platforms
are usually affected by various disturbances, such as waves, currents, winds, ice
and earthquakes. As an efficient method to cope with parameter uncertainties and

unmodeled dynamics, sliding mode control strategies [97] are recently introduced to
control vibration of offshore platforms. In [98], an optimal sliding mode controller
with a specified decay rate is proposed to control wave-induced vibration of an
offshore platform with an AMD mechanism. Based on a linear transformation, an
offshore platform system is decomposed into two subsystems. The velocity of an
AMD is considered as a virtual control for the first subsystem, and an optimal virtual
controller is derived. Then an optimal sliding mode controller with a specified decay
rate is proposed for the original system. It is found that response amplitudes of
displacement and velocity of the offshore platform with an optimal sliding mode
controller are smaller than those with a discrete FFOC [81]. Moreover, control force
required by the former is smaller than that by the latter.
For an offshore steel jacket platform with an active TMD mechanism [65, 72],
robust integral sliding mode control schemes for the offshore platform subject to
hydrodynamic forces and parameter perturbations are developed [99]. Simulation
results show that the performance of the offshore platform with integral sliding
mode controllers is better than the one with a nonlinear controller [65] and a
dynamic output feedback controller [100]. In [101], by taking into account the
uncertainties not only on the natural frequency and the damping ratio of the offshore
platform and the TMD but also on the damping and stiffness of the TMD, a new
uncertain dynamic model for the offshore platform is established. Based on the
proposed dynamic model, sliding mode controllers with mixed current and delayed
states are presented. In [102], by combining sliding mode control with an adaptive
control algorithm and a wavelet support vector machine method, an adaptive
integral sliding mode controller is developed to improve the control performance
of the offshore platform. In [103], an adaptive output feedback integral sliding
mode control schemes are investigated. Note that in [65, 72, 99, 101–103], only
hydrodynamic forces and parametric uncertainties are considered, while external
loading is ignored. Fortunately, in [90], by taking into account external disturbance
acting on the first two vibration modes and parametric uncertainties, robust sliding
mode H∞ controller is designed to reduce oscillation amplitudes of the offshore
platform. It is found through simulation results that compared with a traditional
H∞ controller and a sliding mode controller, a sliding mode H∞ controller requires
much less control force; and oscillation amplitudes of the offshore platform with a
sliding mode H∞ controller are less than those with a sliding mode controller.
The sliding mode control schemes aforementioned are mainly based on two
simplified dynamic models [65, 72, 79] of offshore steel jacket platforms, and the
effectiveness of the designed controllers is demonstrated only by simulation results.
From the control implementation point of view, vibration experiments even field
tests for offshore platforms are necessary. It should be mentioned that, because of
high efficiency, simplicity, and robustness, sliding mode control is also a potential
alternative to vibration reduction for other offshore platforms except for steel jacket
platforms.
12 1 Introduction
1.3.4 Delayed Feedback Control
Under some circumstances, by properly introducing time-delays into control chan-

nel, control performance of some practical systems can be improved [104–106].
Bearing that into mind, several delayed feedback control strategies are introduced
to mitigate vibration and improve the performance of offshore platforms. In the
following, a brief review is presented on this issue.
In [100], by artificially introducing a constant time-delay into the control
channel, a delayed dynamic output feedback controller (DOFC) is designed for an
offshore steel jacket platform with an active TMD mechanism, and the effect of
the introduced time-delay on dynamic output feedback control is investigated. It
is shown through simulation results that the gain of the delayed DOFC is smaller
than the one of a conventional DOFC in the sense of Euclidean norm, while internal
oscillation amplitudes of the offshore platform can be significantly reduced by a
delayed DOFC. Thus, it is clear that when properly introduced, a time-delay can
improve control performance of the offshore platform. However, some limitation
still remains in [100]. In fact, the designed controller requires a large control force.
In addition, the controller is effective only if the introduced time-delay is less than
0.11 s, which is quite small from an implementation point of view. To solve this
problem, in [101], a robust sliding mode controller using mixed current and delayed
states is developed, and the effects of time-delays on sliding mode control for an
offshore steel jacket platform are investigated. It is observed that the proposed
controller is more effective in both improving the control performance and reducing
control force of the offshore platform than some existing ones, such as a delay-
free sliding mode controller [99], a dynamic output feedback controller [100], and
a nonlinear controller [65]. Furthermore, it is found that the introduced time-delay
in this scheme can take values in different ranges, while the corresponding control
performance of the offshore platform is almost at the same level. Recently, by using
both the current and delayed signals [101], delayed non-fragile H∞ control [91],
delayed adaptive sliding mode control [102], and delayed adaptive output feedback
sliding mode control [103] are designed, and the performance of the proposed
controllers is analyzed based on simulation results. In [107], a time-varying delay is
intentionally introduced into the control channel, and a delayed robust sliding mode
H∞ controller is developed to reduce internal oscillations of an offshore platform.
The positive effect of the introduced time-varying delay on robust sliding mode H∞
control for the system is investigated. It is found that control force required by the
proposed controller is smaller than the one by a robust sliding mode controller and
a robust sliding mode H∞ controller [90].
On the other hand, by only using delayed signals, pure delayed feedback
control schemes are developed for offshore platforms. For example, for an offshore
steel jacket platform with an AMD mechanism, a pure delayed H∞ controller is
developed to attenuate wave-induced vibration [89]. It is found that, compared with
the FFOC [81], both vibration amplitudes of the offshore platform and required
control force with the pure delayed H∞ controller are smaller than those with an
FFOC. It should be mentioned that vibration amplitudes of the offshore steel jacket
platform with the pure delayed H∞ controller are at the same level as the one with
a delay-free H∞ controller, while required control force by the former is much
smaller than the one by the latter. Similar results can also be found in [108], where a
pure delayed state feedback non-fragile control scheme is investigated for offshore
platforms with a TMD mechanism and controller perturbations. It is observed that
by choosing proper time-delays, vibration amplitudes of the offshore platforms with
a pure delayed control scheme are the same as or smaller than the ones with a delay-
free control scheme, while required control force by the delayed control scheme is
smaller than that by the delay-free control scheme.
Under the above control schemes, state/output feedback controllers are designed
in the continuous-time domain. When a continuous-time controller for an offshore
platform is implemented in practice, control signals are usually transmitted in a
digital form, which results in a sampled-data system. Hence, some sampled-data
control schemes are introduced to control offshore platforms. By taking input time-
varying delays, actuator faults, and linear fractional uncertainties into account, a
robust fault-tolerant sampled-data H∞ controller is developed to suppress wave-
induced vibration of an offshore steel jacket platform [109]. For an offshore
steel jacket platform subject to hydrodynamic forces, robust sampled-data control
schemes are presented to attenuate vibration of the platform with actuator faults and
parametric perturbations [110–113]. Simulation results indicates that the sampled-
data control schemes can significantly reduce wave-induced vibration and thereby
improve the control performance of the offshore platforms. Furthermore, it is found
that, compared with continuous-time controllers, sampled-data controllers may take
less control cost, while vibration amplitudes of offshore platforms can be reduced
to the similar level.
It is true that intentionally introducing proper time-delays can reduce vibration
amplitudes and required control force of an offshore platform. However, some
challenging issues need to be addressed. For instance, delayed feedback control
schemes are mainly based on simplified dynamic models of offshore steel jacket
platforms. As a result, only the first vibration mode or the first two vibration modes
are considered, while other higher vibration modes are ignored. Moreover, how
to choose a proper time-delay with positive effects on offshore platforms is still
challenging.
1.3.5 Network-Based Control
Network-based control provides an advanced remote control strategy, where control

components, i.e., physical plants, sensors, controllers, and actuators, are connected
through a communication network. Network-based control has several advantages
including low control costs, ease of installation and maintenance, and high reliability
and thus has gained much attention [114–119]. On the other hand, offshore
platforms are generally located far away from land and always affected by com-
14 1 Introduction
plicated and harsh ocean environmental loads. Thus, network-based control paves
an effective way to lower control costs and simplify installation and maintenance
of an offshore platform, while the safety of staff on the platform can be ensured. In
the recent years, several network-based modeling and active network-based control
schemes for offshore platforms are developed.
Based on an offshore platform with an active TMD mechanism [65, 72] or an
AMD mechanism [79], network-based dynamic modeling and controller design are
presented in [120] and [123], respectively. By inserting a communication network
over the offshore platform, a network-based model of the offshore platform is
established. Then a network-based state feedback controller is designed to suppress
the amplitudes of the offshore platform [120]. Under this scheme, the corresponding
closed-loop system of the offshore platform is modeled as a system with an interval
time-varying delay [121, 122]. Simulation results show that both the oscillation
amplitudes of the offshore platform and the required control force with a network-
based state feedback controller are smaller than those with a nonlinear controller
[65] and a dynamic output feedback controller [100]. Moreover, oscillation ampli-
tudes of the offshore platform with a network-based feedback controller are almost
the same as those with an integral sliding mode controller [99], while the required
control force is smaller than that required by an integral sliding mode controller.
In [123], a network-based control model is introduced for an offshore platform
with an AMD mechanism in the presence of actuator faults, where an event-
triggered mechanism is utilized to save the limited resources of the communication
networks [124, 125]. Based on a network-based dynamic model, an event-triggering
H∞ reliable controller is designed by employing the Lyapunov-Krasovskii func-
tional approach. It is found that for possible actuator failures, a network-based
controller is capable of guaranteeing the stability of the offshore platform. Com-
pared with a traditional H∞ controller, a network-based controller can suppress
vibration of the offshore platform to almost the same level while requiring less
control costs. Furthermore, with the event-triggered H∞ controller, limited network
resources can be saved significantly.
As mentioned above, network-based control can improve the control perfor-
mance of offshore platforms. However, using communication networks unavoidably
leads to some unfavorable factors, such as network-induced delays, packet dropouts,
quantization errors, and network congestion. By taking network-induced delays
into account, network-based control for offshore platforms is studied in [120]
and [123]. However, sustained attention to network-based control for offshore
platforms is still worthy to be paid. Specifically, by taking one or more factors
aforementioned into consideration, it is an interesting research topic to explore a
network-based dynamic model and develop effective network-based controllers for
floating offshore platforms in ocean environments.
1.4 Book Outline 15
1.4 Book Outline
The book is organized as follows: Chap. 1 presents an overview of recent devel-

opments of vibration control of offshore platforms; Chap. 2 gives some dynamic
models of the offshore platforms and several mathematical lemmas used in this
book; Chaps. 3 and 4 devote to optimal tracking control schemes and integral
sliding mode H∞ control schemes, respectively; Chaps. 5, 6, and 7 focus on delayed
sliding mode control scheme, delayed non-fragile H∞ control scheme, and delayed
dynamic output feedback control scheme, respectively; and Chaps. 8 and 9 aim at
dealing with network-based modeling and network-based control issues of offshore
platforms.
In Chap. 3, optimal tracking control schemes for offshore platforms with AMD
mechanisms are proposed, and the existence and uniqueness of optimal tracking
controllers are discussed.
In Chap. 4, integral sliding mode H∞ control schemes for offshore platforms
with active TMD mechanisms are developed.
In Chap. 5, by intentionally introducing a proper time-delay into the control
channel, a novel sliding mode control scheme with mixed current and delayed states
is proposed. A computational algorithm to solve gain matrices of delayed sliding
mode controllers is presented.
In Chap. 6, for an offshore steel jacket platform subject to self-excited nonlinear
hydrodynamic force, external disturbance, and controller gain perturbations, a
delayed robust non-fragile H∞ control scheme is developed. The positive effects
of the time-delays on the non-fragile H∞ control for the offshore platform are
investigated.
In Chap. 7, a dynamic output feedback control scheme and a delayed dynamic
output feedback control scheme are presented for an offshore steel jacket platform
with TMD mechanisms. The effects of intentionally introduced time-delays on
dynamic output feedback control of the offshore platform are investigated.
In Chap. 8, a network-based dynamic model of an offshore steel jacket platform
is established and a network-based state feedback control scheme for the offshore
platform is developed.
In Chap. 9, an event-triggered H∞ reliable control problem for an offshore
platform under external wave force and actuator faults is investigated. An event-
triggering approach is presented to deal with the issue on limited network bandwidth
in the controller design.
Chapter 2
Dynamic Models of Offshore Platforms
In this chapter, two dynamic models of offshore platforms and several required
lemmas are introduced for investigating active control strategies in this book. In
the first dynamic model, only the first dominant vibration mode of an offshore steel
jacket platform with an AMD mechanism is taken into account [79]. This model
is utilized to design active controllers to attenuate wave-induced vibration of the
offshore platform. In the second dynamic model, the first and the second vibration
modes of an offshore steel jacket platform subject to an active TMD mechanism
are considered [65, 72]. By considering parametric perturbations of the system and
external disturbance, several uncertain nonlinear models for the offshore platform
are developed. Such models are used to design active controllers to reduce vibration
amplitudes of the offshore platform subject to self-excited hydrodynamic forces
and/or external disturbance.
2.1 Model of an Offshore Platform with AMD Mechanisms
Consider a simplified offshore steel jacket platform with an AMD mechanism

showed in Fig. 2.1 [79], where only the first dominant vibration mode is considered.
The motion equation of the offshore platform subject to external wave forces is
given by
⎧
⎨ m1 z̈1 (t) = −(m1 ω12 + m2 ω22 )z1 (t) + m2 ω22 z2 (t) + 2ξ2 ω2 m2 ż2 (t)
−2(m1 ξ1 ω1 + m2 ξ2 ω2 )ż1 (t) + f (t) − u(t) (2.1)
⎩
m2 z̈2 (t) = m2 ω22 [z1 (t) − z2 (t)] + 2m2 ξ2 ω2 [ż1 (t) − ż2 (t)] + u(t)

https://doi.org/10.1007/978-981-13-2986-9_2
18 2 Dynamic Models of Offshore Platforms
Fig. 2.1 An offshore steel jacket structure with an AMD [79]
where m1 , ω1 , and ξ1 are the modal mass, frequency, and damping ratio of
the offshore structure, respectively; m2 , ω2 , and ξ2 are the mass, frequency, and
damping ratio of the AMD, respectively; z1 (t) denotes the corresponding modal
coordinate which refers to the deck motion of the offshore structure, and z2 (t) is
displacement of the AMD; u(t) is the active control force, and f (t) is the external
wave forces.
Let
x1 (t) = z1 (t), x2 (t) = z2 (t), x3 (t) = ż1 (t), x4 (t) = ż2 (t) (2.2)
and denote
x(t) = [x1 (t) x2 (t) x3 (t) x4 (t)]T
Then, the motion equation (2.1) can be written as
ẋ(t) = Ax(t) + Bu(t) + Df (t), x(0) = x0 (2.3)

2.1 Model of an Offshore Platform with AMD Mechanisms 19
where x0 is the initial state vector, and

⎧ ⎡ ⎤
⎪
⎪ 0 0 1 0
⎪
⎪ ⎢ 0 0 ⎥
⎪
⎪ 0 1
⎪
⎪ A=⎢
⎣ ā31 ā32 ā33
⎥
⎪
⎪
⎪ ā34 ⎦
⎨
ω2 −ω22 2ξ2 ω2 −2ξ2 ω2
2 (2.4)
⎪
⎪ 1 1 T
⎪
⎪ B= 0 0 −
⎪
⎪ m1 m2
⎪
⎪
⎪
⎪ 1 T
⎪
⎩D = 0 0 0
m1
with
m2 ω22 m2 ω22
ā31 = −ω12 − , ā32 =
m1 m1
2m2 ξ2 ω2 2m2 ξ2 ω2
ā33 = −2ξ1 ω1 − , ā34 =
m1 m1
It is assumed that the wave force f (t) acting on the offshore structure comes
from the direction of x-axis, and the wave propagation is unidirectional [79, 82].
The power spectral density function of the wave elevation η(t) is given by the Joint
North Sea Wave Project (JONSWAP) wave spectrum as
−4
5Hs2 ω0 5 ω
Sη (ω) = exp −1.25 γ̄ β (2.5)
16ω0 ω ω0
where Hs denotes the significant wave height, γ̄ ∈ [0, Hs ] is the peakedness

coefficient, ω0 is the peak frequency, and ω is the wave frequency; and

ω − ω0 2
β = exp − √
2θ ω0
with θ = 0.07 for ω ≤ ω0 and θ = 0.09 for ω > ω0 .

Suppose that z represents the vertical coordinate, d0 represents the water depth,
and m̄ denotes the wave number satisfying
ω2 = g m̄ tanh(m̄d0 ) (2.6)
where g denotes the gravitational acceleration. Then based on the linear wave
theory, for the wave elevation η(t) with wave frequency ω, the water particle
velocity υ(z, t), the acceleration υ̇(z, t), and the standard deviation συ (z) of the
velocity at location z can be written as
υ(z, t) = Tυη (ω, z)η(t) (2.7)

υ̇(z, t) = Tυ̇η (ω, z)η(t) (2.8)
ω 1/2
σv (z) = |Tυ̇η (ω, z)|2 Sη (ω)dω (2.9)
0
where
cosh(mz) cosh(mz)
Tυη (ω, z) = ω , Tυ̇η (ω, z) = −j ω2 (2.10)
sinh(md0 ) sinh(md0 )
√
with j = −1.
Based on the linearized Morison equation [126, 127], along the cylindrical
structural members, the physical horizontal wave force p(z, t) per unit length is
in the form

1 8 1
p(z, t) = ρCd D̃ συ (z)υ(z, t) + ρπ Cm D̃ 2 υ̇(z, t) (2.11)
2 π 4
where ρ is the fluid density, D̃ represents the cylinder diameter, Cd is drag

coefficient, and Cm is the inertia coefficient. Then, the total force f (t) on the
structure can be formulated as
d0
f (t) = p(z, t)φ(z)dz (2.12)
0
where φ(z) is the shape function as

πz
φ(z) = 1 − cos , 0≤z≤L (2.13)
2L
with L is the length of an offshore structure.
Based on (2.12), the wave force f (t) on the offshore structure can be computed
numerically. Specifically, for simplicity, representing the wave elevation η(t) as [81]

n
η(t) = ηj (t) (2.14)
j =1
where n is a specified positive integer and ηj (t) denotes the j th component of the
wave elevation with the form as
ηj (t) = Aj cos(−ωj t + ςj ) (2.15)

2.1 Model of an Offshore Platform with AMD Mechanisms 21
with Aj is the wave amplitude, ωj is the wave frequency, and ςj is the random
phase angle uniformly distributed between 0 and 2π .
Furthermore, denote
T
v(t) = η1 (t) η2 (t) · · · ηn (t)
T (2.16)
w(t) = v T (t) v̇ T (t)
and
⎧
⎨ G̃ = −diag{ω12 , ω22 , · · · , ωn2 }
n
(2.17)
⎩ H̃ = [1 1 · · · 1] T (ωj )
j =1
where

d0 1 8 1
T (ωj ) = ρCd D̃ συ (z)Tυη (ωj , z) + ρπ Cm D̃ Tυ̇η (ωj , z) φ(z)dz
2
0 2 π 4
Then, from (2.5), (2.11), and (2.12), the wave force f (t) can be modeled as the
output of a linear exogenous system as [81]
ẇ(t) = Gw(t), f (t) = H w(t) (2.18)
where

0̃ I˜
G= , H = H̃ I˜ 0̃ (2.19)
G̃ 0̃
with 0̃ and I˜ represent the n×n null matrix and the n×n identity matrix, respectively,
and matrices G̃ and H̃ are given by (2.17).
It should be specified that as an approximate representation of wave force, it can
be seen from (2.19) that the exogenous system (2.18) is stable but not asymptotically
stable, i.e., for any pole λ of the exogenous system, the following is true:
Re(λ) = 0, ∀λ ∈ σ (G) (2.20)
where σ (·) denotes the spectrum of a matrix. Specifically, it should be pointed herein
that the condition (2.20) is necessary for designing an optimal tracking controller in
Chap. 3.
2.2 Model of an Offshore Platform with Active TMD

Mechanisms
Consider an offshore steel jacket platform with three floors shown in Fig. 2.2 [11,
65, 72]. The platform is assembled from a concrete deck and 13 cylindrical steel
tube members with the dimensions reported in [11]. An active TMD mechanism,
which consists of a small mass, a spring and a viscous damper, is mounted on the
top of the offshore platform and is excited by a hydraulic servo mechanism. The
motion of the platform and the operation of the hydraulic servo influence the motion
of the damper. The hydraulic servo is driven by the active control force to reduce
vibration. The offshore platform is exposed to hydrodynamic force, which induces
a self-excited load term [65]. For simplicity, a monochromatic wave acting on the
offshore platform is considered and shown in Fig. 2.2, where h is the water depth,
H the wave height, λ the wave length, and Uow the current velocity at the water
surface. For details, one can refer to [11]. Since the first two modes of vibration
are the most dominant for controller design [65, 72, 79], in the sequel, we consider
the motions of the first two modes. When taking the parameter uncertainties of the
first and the second modes of the vibration and the coupled TMD, and external
disturbances acting on the two modes into consideration, the motion equations of
the first two modes of the vibration with the TMD can be described by
Fig. 2.2 An offshore steel jacket structure with a TMD [11, 65, 72]
2.2 Model of an Offshore Platform with Active TMD Mechanisms 23
⎧
⎪
⎪ z̈1 (t) = − [ω1 + Δω1 (t)]2 z1 (t) − 2[ξ1 + Δξ1 (t)][ω1 + Δω1 (t)]ż1 (t)
⎪
⎪
⎪
⎪ − φ1 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + φ1 [ζ1 (t) − u(t)]
⎪
⎪
⎪
⎪
⎪
⎪ − φ1 [CT + ΔCT (t)][φ1 ż1 (t) + φ2 ż2 (t)] + φ1 [KT + ΔKT (t)]zT (t)
⎪
⎪
⎪
⎪
⎪
⎪ + φ1 [CT + ΔCT (t)]żT (t) + f1 (z1 (t), z2 (t), t) + f2 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨ z̈2 (t) = − [ω2 + Δω2 (t)] z2 (t) − 2[ξ2 + Δξ2 (t)][ω2 + Δω2 (t)]ż2 (t)
2
⎪ − φ2 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + φ2 [ζ2 (t) − u(t)]

⎪
⎪
⎪
⎪
⎪
⎪ − φ2 [CT + ΔCT (t)][φ1 ż1 (t) + φ2 ż2 (t)] + φ2 [KT + ΔKT (t)]zT (t)
⎪
⎪
⎪
⎪
⎪
⎪ + φ2 [CT + ΔCT (t)]żT (t) + f3 (z1 (t), z2 (t), t) + f4 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪ 1
⎪
⎪ z̈T (t) = − [ωT + ΔωT (t)]2 [zT (t) − φ1 z1 (t) − φ2 z2 (t)] + u(t)
⎪
⎪ m
⎪
⎪
T
⎩
− 2[ξT + ΔξT (t)][ωT + ΔωT (t)][żT (t) − φ1 ż1 (t) − φ2 ż2 (t)]
(2.21)
where z1 (t) and z2 (t) are generalized coordinates of the first and the second
vibration modes, respectively; zT (t) is the horizontal displacement of the TMD;
ω1 and ω2 are nominal natural frequencies of the first and the second vibration
modes, respectively; Δω1 (t) and Δω2 (t) represent the perturbations with respect to
nominal parameters ω1 and ω2 , respectively; ξ1 and ξ2 are nominal damping ratios of
the first and the second vibration modes, respectively; Δξ1 (t) and Δξ2 (t) represent
the perturbations with respect to nominal parameters ξ1 and ξ2 , respectively; φ1
and φ2 are mode shapes vectors of the first and the second vibration modes,
respectively; ωT and ξT are the nominal natural frequency and damping ratio of
the TMD, respectively; ΔωT (t) and ΔξT (t) represent perturbations with respect
to nominal parameters ωT and ξT , respectively; CT , mT and KT are the nominal
damping, mass and stiffness of the TMD, respectively; ΔCT (t) and ΔKT (t) rep-
resent perturbations with respect to nominal parameters CT and KT , respectively;
fi (z1 (t), z2 (t), t), i = 1, 2, 3, 4 are the self-excited wave force terms; ζ1 (t) and
ζ2 (t) represent the external disturbances acting on the first and the second vibration
modes, respectively; and u(t) is the active control force of the system.
Suppose that time-varying perturbations ΔKT (t), ΔCT (t), and Δξi (t) are
described as
⎧
⎨ ΔKT (t) = K̂T · ΔK̃T (t)
ΔCT (t) = ĈT · ΔC̃T (t) (2.22)
⎩
Δξi (t) = ξ̂i · Δξ̃i (t), i = 1, 2, T
where K̂T , ĈT , and ξ̂i represent the maximum perturbations with respect to the
nominal values of KT , CT , and ξi , respectively, |ΔK̃T (t)| ≤ 1, |ΔC̃T (t)| ≤ 1, and
|Δξ̃i (t)| ≤ 1, i = 1, 2, T .
Notice that K̂T KT leads to

K̂T K̂T
1+ ΔK̃T (t) = 1 + ΔK̃T (t) + o(ΔK̃T (t)) (2.23)
KT 2KT
where o(·) denotes a higher order infinitesimal.
Then, from (2.23) and

KT + ΔKT (t)
ωT + ΔωT (t) = (2.24)
mT
we have
ΔωT (t) = ωT ω̂T · Δω̃T (t) + o(Δω̃T (t)) (2.25)
where
K̂T
ω̂T = , Δω̃T (t) = ΔK̃T (t) (2.26)
2KT
Similarly, the perturbation terms of the natural frequencies of the first and the second
vibration modes are given by
Δωi (t) = ωi ω̂i · Δω̃i (t) + o(Δω̃i (t)), i = 1, 2 (2.27)
where ω̂i is a maximum perturbation with respect to the nominal value ωi and
|Δω̃i (t)| ≤ 1.
Then, by ignoring the higher-order infinitesimal o(Δω̃i (t)) and the second-order
terms Δξ̃i (t)·Δω̃i (t) and Δω̃i2 (t), i = 1, 2, T , it follows from (2.21), (2.22), (2.25),
and (2.27) that the dynamic model (2.21) can be written as
⎧
⎪
⎪ z̈1 (t) = − [ω12 + ω̄1 (t)]z1 (t) − [2ξ1 ω1 + ν̄1 (t)]ż1 (t) + φ1 [ζ1 (t) − u(t)]
⎪
⎪
⎪
⎪ − φ1 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + f1 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪ − φ1 [CT + ΔCT (t)][φ1 ż1 (t) + φ2 ż2 (t)] + f2 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪ + φ1 [KT + ΔKT (t)]zT (t) + φ1 [CT + ΔCT (t)]żT (t)
⎪
⎪
⎪
⎪
⎪
⎪ z̈ (t) = − [ω2 + ω̄ (t)]z (t) − [2ξ ω + ν̄ (t)]ż (t) + φ [ζ (t) − u(t)]
⎪
⎨ 2 2 2 2 2 2 2 2 2 2
⎪ − φ2 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + f3 (z1 (t), z2 (t), t)

⎪
⎪
⎪
⎪
⎪
⎪ − φ2 [CT + ΔCT (t)][φ1 ż1 (t) + φ2 ż2 (t)] + f4 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪ + φ2 [KT + ΔKT (t)]zT (t) + φ2 [CT + ΔCT (t)]żT (t)
⎪
⎪
⎪
⎪
⎪
⎪ 1
⎪
⎪ z̈T (t) = − [ωT2 (t) + ω̄T (t)][zT (t) − φ1 z1 (t) − φ2 z2 (t)] + u(t)
⎪
⎪ mT
⎪
⎪
⎩
− [2ξT ωT + ν̄T (t)][żT (t) − φ1 ż1 (t) − φ2 ż2 (t)]
(2.28)
where
ν̄i (t) = 2ωi [ξi ω̂i · Δω̃i (t) + ξ̂i · Δξ̃i (t)], i = 1, 2, T (2.29)
and
ω̄i (t) = 2ωi2 ω̂i · Δω̃i (t), i = 1, 2, T (2.30)
Denote
⎧
⎪
⎪ =
0 0
⎨ ΔA0 (t)
ΔKT ΔCT
(2.31)
⎪
⎪ 0 0
⎩ ΔAi (t) = , i = 1, 2, T
ω̄i (t) ν̄i (t)
Then, introduce a time-varying 3 × 3 block matrix ΔA(t) as

ΔA(t) = Aij (t) 3×3 (2.32)
where
A11 (t) = −φ12 · ΔA0 (t) − ΔA1 (t), A12 (t) = −φ1 φ2 · ΔA0 (t)
A22 (t) = −φ22 · ΔA0 (t) − ΔA2 (t), A13 (t) = φ1 · ΔA0 (t)
A21 (t) = A12 (t), A23 (t) = φ2 · ΔA0 (t)
A31 (t) = φ1 · ΔAT (t), A32 (t) = φ2 · ΔAT (t)
A33 (t) = −ΔAT (t)
Note that

ΔA0 (t) = Ē0 · diag{Δω̃T (t), ΔC̃T (t)}
(2.33)
ΔAi (t) = Ēi · diag{Δω̃i (t), Δξ̃i (t)} · H̄i , i = 1, 2, T
where
⎧
⎪
⎪ 0 0 0 0
⎨ Ē0 = , Ēi =
K̂ Ĉ ω̂i ξ̂i
T2 T (2.34)
⎪
⎪ 2ωi 2ξi ωi
⎩ H̄i = , i = 1, 2, T
0 2ωi
Denote
⎧
⎪
⎪ Ẽ = diag{Ē1 , Ē2 , ĒT }, Ẽ0 = Φ0 ⊗ Ē0
⎪
⎪ ⎡ ⎤
⎪
⎨ −H̄1 0 0
H̃ = ⎣ 0 −H̄2 0 ⎦ (2.35)
⎪
⎪
⎪
⎪ φ1 H̄T φ2 H̄T −H̄T
⎪
⎩ H̃ = diag{−I, −I, I }
0
where ⊗ represents the Kronecker product of matrices, and

⎡ ⎤
φ12 φ1 φ2 φ1
Φ0 = ⎣ φ1 φ2 φ22 φ2 ⎦
0 0 0
Let
⎧
⎪
⎨ M̃ = Ẽ Ẽ0
T T
Ñ = H̃ H̃0T (2.36)
⎪
⎩
F̃ (t) = diag{F̃1 (t), F̃0 (t)}
where

F̃1 (t) = diag{Δω̃1 (t), Δξ̃1 (t), Δω̃2 (t), Δξ̃2 (t), Δω̃T (t), Δξ̃T (t)}
F̃0 (t) = diag{Δω̃T (t), ΔC̃T (t), Δω̃T (t), ΔC̃T (t), Δω̃T (t), ΔC̃T (t)}
(2.37)
Then, the perturbation matrix ΔA(t) in (2.32) can be written as
ΔA(t) = M̃ F̃ (t)Ñ (2.38)
where the time-varying parameter perturbation F̃ (t) satisfies
F̃ T (t)F̃ (t) ≤ I, ∀t ≥ 0 (2.39)
Denote

x1 (t) = z1 (t), x2 (t) = ż1 (t), x3 (t) = z2 (t)
(2.40)
x4 (t) = ż2 (t), x5 (t) = zT (t), x6 (t) = żT (t)
and
x(t) = [ x1 (t) x2 (t) x3 (t) x4 (t) x5 (t) x6 (t) ]T

f1 (z1 (t), z2 (t), t) + f2 (z1 (t), z2 (t), t)
f (x, t) =
f3 (z1 (t), z2 (t), t) + f4 (z1 (t), z2 (t), t)

ζ (t)
ζ (t) = 1
ζ2 (t)
Then the dynamic model (2.28) of the offshore platform can be written as
ẋ(t) = [A + M̃ F̃ (t)Ñ]x(t) + Bu(t) + Df (x, t) + D0 ζ (t), x(0) = x0

(2.41)
where x0 is the initial value, M̃, F̃ (t), and Ñ are denoted by (2.36), and the matrices
A, B, D, and D0 are given as
⎧ ⎡ ⎤
⎪
⎪
0 1 0 0 0 0
⎪
⎪ ⎢a a a a a a ⎥
⎪
⎪ ⎢ 21 22 23 24 25 26 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 0 0 1 0 0 ⎥
⎪
⎪ A=⎢ ⎥
⎪
⎪ ⎢ a41 a42 a43 a44 a45 a46 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎣ 0 0 0 0 0 1 ⎦
⎪
⎪
⎨ a a a a a a
61 62 63 64 65 66 T (2.42)
⎪
⎪ 1
⎪
⎪ B = 0 −φ1 0 −φ2 0
⎪
⎪ mT
⎪
⎪ T
⎪
⎪ 0 1 0 0 0 0
⎪
⎪ D=
⎪
⎪
⎪
⎪
000100
T
⎪
⎪
⎪
⎪ 0 φ1 0 0 0 0
⎩ D0 =
0 0 0 φ2 0 0
with
a21 = −ω12 − KT φ12 , a22 = −2ξ1 ω1 − CT φ12

a23 = −KT φ1 φ2 , a24 = −CT φ1 φ2
a25 = φ 1 KT , a26 = φ1 CT
a41 = −KT φ1 φ2 , a42 = −CT φ1 φ2
a43 = −ω22 − KT φ22 , a44 = −2ξ2 ω2 − CT φ22
a45 = φ 2 KT , a46 = φ2 CT
a61 = ωT2 φ1 , a62 = 2ξT ωT φ1
a63 = ωT2 φ2 , a64 = 2ξT ωT φ2
a65 = −ωT2 , a66 = −2ξT ωT .
It is assumed that in (2.41), the external disturbance term ζ (t) ∈ L2 [0, ∞]. As
shown in [65, 72], the self-excited wave force vector f (x, t) is uniformly bounded
and satisfies
f (x, t) ≤ μ x (2.43)
where μ is a positive scalar.

Remark 2.1 It should be pointed out that an approximate uncertain dynamic
model (2.41) for the offshore steel jacket platform is obtained for (2.21), where
uncertainties on natural frequencies, on damping ratios of the first two vibration
modes and the TMD, on damping and stiffness of the TMD, and the external
disturbance are taken into account. In fact, the dynamic model (2.41) is more general
than the existing models in [65, 72, 90, 99] and [101].
As special cases of the model (2.41), several dynamic models of the offshore
platform are listed as follows.
In (2.21), if the external disturbance is not considered, the uncertain dynamic
model (2.41) reduces to the one in [101] as
ẋ(t) = [A + M̃ F̃ (t)Ñ]x(t) + Bu(t) + Df (x, t), x(0) = x0 (2.44)
Further, if setting ΔKT (t) ≡ 0 and ΔCT (t) ≡ 0, Δω̃i (t) ≡ 0 and Δξ̃i (t) ≡ 0, i =
1, 2, T , the dynamic model (2.44) reduces to the ones in [65, 72] and [100] as
ẋ(t) = Ax(t) + Bu(t) + Df (x, t), x(0) = x0 (2.45)
In (2.21), if the uncertainties on damping and stiffness of the TMD are not
taken into account, and the perturbation terms Δω1 (t), Δω2 (t), and ΔωT (t) of
the nominal natural frequencies of the first two vibration modes and the TMD are
approximated by the additive uncertainty forms, the terms ν̄i (t) in (2.29) and ω̄i (t)
in (2.30) can be written as

ν̄i (t) = 2ξi ω̂i · Δω̃i (t) + 2ωi ξ̂i · Δξ̃i (t)
(2.46)
ω̄i (t) = 2ωi ω̂i · Δω̃i (t), i = 1, 2, T
In this case, the uncertain dynamic model (2.41) reduces to the one in [90] and [91]
as
ẋ(t) = [A + M̂ F̂ (t)N̂]x(t) + Bu(t) + Df (x, t) + D0 ζ (t), x(0) = x0

(2.47)
where
⎧
⎪
⎪ M̂ = Ẽ
⎨ ⎡ ⎤
−Ĥ1 0 0
(2.48)
⎪
⎪ N̂ = ⎣ 0 −Ĥ2 0 ⎦
⎩
φ1 ĤT φ2 ĤT −ĤT
and F̂ (t) = F̃1 (t) satisfying F̂ T (t)F̂ (t) ≤ I, ∀t ≥ 0, and Ẽ and F̃1 (t) are given
by (2.35) and (2.37), respectively, and

2ωi 2ξi
Ĥi = , i = 1, 2, T
0 2ωi
Specifically, in (2.47), setting ζ (t) ≡ 0 yields
ẋ(t) = [A + M̂ F̂ (t)N̂]x(t) + Bu(t) + Df (x, t), x(0) = x0 (2.49)

2.3 Some Related Mathematical Lemmas 29
If the perturbations of the systematic parameters are not considered, the nominal
form of the offshore platform systems (2.41) and (2.47) is derived as
ẋ(t) = Ax(t) + Bu(t) + Df (x, t) + D0 ζ (t), x(0) = x0 (2.50)
2.3 Some Related Mathematical Lemmas
In this section, some main lemmas are provided to obtain the main results of this
book. The proof of these lemmas can be found in the literature.

S11 S12
Lemma 2.1 (Schur complement) For a symmetric matrix S = , where
∗ S22
S11 ∈ Rn×n . Then the following conditions are equivalent:
1. S < 0;
T S −1 S < 0;
2. S11 < 0, S22 − S12 11 12
−1 T
3. S22 < 0, S11 − S12 S22 S12 < 0.
Lemma 2.2 ([128]) Given a symmetric matrix Y and matrices H , F (t), E of
appropriate dimensions with F T (t)F (t) ≤ I , then
Y + H F (t)E + E T F T (t)H T < 0
if and only if there exists a scalar ε > 0 such that
Y + εH H T + ε−1 E T E < 0
Lemma 2.3 ([129]) For matrices Ã ∈ Rn×n , B̃ ∈ Rm×m , Γ ∈ Rn×m , and X ∈

Rn×m , the matrix equation
ÃX + XB̃ + Γ = 0 (2.51)
has a unique solution X if and only if
μi + νj = 0, i = 1, 2, · · · , n, j = 1, 2, · · · , m, (2.52)
where μi ∈ σ (Ã) and νj ∈ σ (B̃).

Lemma 2.4 (Projection Theorem [130]) Let W = W T ∈ Rn×n , U ∈ Rn×m , and
V ∈ Rk×n be given matrices. Then inequality
W + U GV T + (U GV T )T < 0 (2.53)
is solvable for matrix G with appropriate dimensions if and only if the following
holds
U⊥T W U⊥ < 0, V⊥T W V⊥ < 0 (2.54)
where U⊥ is denoted as the orthogonal complement of U .

Lemma 2.5 ([100]) There exists a positive-definite symmetric matrix P ∈ Rm×m
satisfying

X N −1 Y M
P = , P = (2.55)
N T Z1 M T Z2
if and only if X − Y −1 ≥ 0, where X, Y ∈ Rr1 ×r1 , N, M ∈ Rr1 ×(m−r1 ) , and

Z1 , Z2 ∈ R(m−r1 )×(m−r1 ) .
Lemma 2.6 ([131]) For any constant matrix R ∈ Rn×n , R = R T > 0, a scalar
h > 0 and a vector-valued function ẋ : [t − h, t] → Rn such that the following
integration is well defined, then
t T
x(t) −R R x(t)
−h ẋ (s)R ẋ(s)ds ≤
T
(2.56)
t−h x(t − h) ∗ −R x(t − h)
Lemma 2.7 ([132]) Let x(t) ∈ R n be a vector-valued function with first-order

continuous-derivative entries. Then, the following integral inequality holds for
any matrices X, M1 , M2 ∈ R n×n and Z ∈ R 2n×2n , and a scalar function
h := h(t) ≥ 0:
t
− ẋ T (s)Xẋ(s)ds ≤ ξ T (t)Mξ(t) + hξ T (t)Zξ(t) (2.57)
t−h
where

M1T + M1 −M1T + M2
M :=
∗ −M T − M
2 2
x(t) XY
ξ(t) := , ≥0
x(t − h) ∗ Z

with Y := M1 M2
Lemma 2.8 ([133]) For any constant matrix R ∈ Rn×n , R = R T , matrix S ∈
Rn×n , R = [ R∗ RS ] ≥ 0, scalars 0 ≤ τ1 ≤ τ (t) ≤ τ2 , and a vector function
ẋ : [−τ2 , −τ1 ] → Rn such that the following integration is well defined, it holds
that
2.3 Some Related Mathematical Lemmas 31
t−τ1
4
− (τ2 − τ1 ) ẋ T (s)R ẋ(s)ds ≤ − ∇k (2.58)
t−τ2 k=1
where
∇1 = [x T (t − τ (t)) − x T (t − τ2 )]R[x(t − τ (t)) − x(t − τ2 )]

∇2 = [x T (t − τ1 ) − x T (t − τ (t))]R[x(t − τ1 ) − x(t − τ (t))]
∇3 = [x T (t − τ (t)) − x T (t − τ2 )]S[x(t − τ1 ) − x(t − τ (t))]
∇4 = [x T (t − τ1 ) − x T (t − τ (t))]S T [x(t − τ (t)) − x(t − τ2 )]
Lemma 2.9 ([134]) For any constant matrix R > 0 ∈ Rn×n , a scalar function
γ (t) with 0 < γ (t) ≤ γM , and a vector function
t ė : [−γM , 0] → Rn such that the
following integration is well defined, let t−γ (t) ė(s)ds = Mζ (t) where M ∈ Rn×m
and ζ (t) ∈ Rm . Then the following inequality holds for any matrix N ∈ Rn×m
t
− ėT (s)R ė(s)ds ≤ −ζ T (t) M T N + N T M − γ (t)N T R −1 N ζ (t)
t−γ (t)
(2.59)
Chapter 3
Optimal Tracking Control with
Feedforward Compensation
This chapter presents an optimal tracking control methodology for an offshore steel
jacket platform subject to external wave force. Based on a dynamic model of an
offshore steel jacket platform with an AMD mechanism and a linear exogenous
system model of the external wave force on the offshore platform, an optimal
tracking control scheme with feedforward compensation is proposed to attenuate
wave-induced vibration of the offshore platform. A feedforward and feedback
optimal tracking controller (FFOTC) can be obtained by solving an algebraic Riccati
equation and a Sylvester equation, respectively. It is demonstrated that the wave-
induced vibration amplitudes of the offshore platform under the FFOTC are much
smaller than the ones under the feedback optimal tracking controller (FOTC) and
the feedforward and feedback optimal controller (FFOC). Furthermore, the required
control force under the FFOTC is smaller than the ones under the FOTC and the
FFOC.
This chapter is organized as follows. Section 3.1 presents the linear quadratic
performance index optimal tracking control problem for the offshore steel jacket
platform. Section 3.2 investigates the design process of the feedforward and feed-
back optimal tracking controller. Some numerical examples are given in Sect. 3.3
to demonstrate the effectiveness of the proposed control scheme and its superiority
over some existing methods. Finally, Sect. 3.4 concludes the chapter, and Sect. 3.5
presents a brief note.
3.1 System and Problem Descriptions
Consider an offshore steel jacket platform with an AMD mechanism, which is

shown by Fig. 2.1 [79]. The dynamic model of the offshore platform is presented
by (2.3), where the external disturbance on the offshore platform is modeled as the
output of the exogenous system (2.18).

https://doi.org/10.1007/978-981-13-2986-9_3
34 3 Optimal Tracking Control with Feedforward Compensation
The output equation of system (2.3) is give as
y(t) = Cx(t) (3.1)
where

1000
C= (3.2)
0010
The desired output yr (t) is determined by a given linear system as
ż(t) = Mz(t), yr (t) = N z(t) (3.3)
where M ∈ Rr×r and N ∈ R2×r are constant matrices with appropriate dimensions.
The quadratic average performance index is chosen as
T
1
J = lim eT (t)Qe(t) + uT (t)Ru(t) dt (3.4)
T →∞ 2T 0
where T > 0, Q is a 2 × 2 positive semi-definite matrix, and R > 0 is a scalar, e(t)

is the tracking error as
e(t) = y(t) − yr (t) (3.5)
In what follows, we will design a feedforward and feedback optimal tracking

control law u(t) for system (2.3) such that the output y(t) can asymptotically track
the desired output yr (t), and the performance index (3.4) is minimized.
The following assumptions are needed to obtain the main results of this chapter.
Assumption 1 For the system (2.3) with (3.1), (A, B) is completely controllable,
and the pair (A, C) is completely observable.
Assumption 2 The system (3.3) is asymptotically stable.
Assumption 3 The matrix pair (G, H ) in (2.18) is completely observable.
3.2 Design of Optimal Tracking Controllers
In this section, we present a feedforward and feedback optimal tracking control

scheme to attenuate the wave-induced vibration of the offshore steel jacket platform.
From (3.1), (3.3), and (3.5), one can rewrite the quadratic performance
index (3.4) as
3.2 Design of Optimal Tracking Controllers 35
T
1
J = lim x T C T QCx − 2x T C T QNz + zT N T QNz + uT Ru dt
T →∞ 2T 0
(3.6)
Applying the maximum principle to system (2.3) with the output Eq. (3.1) and
the performance index (3.6) yields the optimal tracking control law satisfying
u(t) = −R −1 B T λ(t) (3.7)
where λ(t) ∈ R4×1 is an adjoint vector satisfying the following two-point boundary
value problem

ẋ(t) = Ax(t) − BR −1 B T λ(t) + Df (t), x(0) = x0
(3.8)
λ̇(t) = −C T QCx(t) + C T QNz(t) − AT λ(t), λ(∞) = 0
To solve the above two-point boundary value problem, let
λ(t) = P1 x(t) + P2 z(t) + P3 w(t) (3.9)
where P1 ∈ R4×4 is the symmetric feedback gain matrix and P2 ∈ R4×r and P3 ∈
R4×2n are feedforward gain matrices to be determined.
Then, the control law (3.7) can be written as
u(t) = −R −1 B T [P1 x(t) + P2 z(t) + P3 w(t)] (3.10)
which indicates that the existence and uniqueness of the optimal tracking control
law (3.10) is equivalent to the existence and uniqueness of the feedback gain matrix
P1 and the feedforward gain matrices P2 and P3 .
Substituting (3.9) into the first equation of (3.8) and noting (2.18) yield
ẋ(t) =(A − BR −1 B T P1 )x(t) − BR −1 B T P2 z(t)

+ (DH − BR −1 B T P3 )w(t) (3.11)
On the one hand, from (3.3), (3.9), and (3.11), we obtain
λ̇(t) =(P1 A − P1 BR −1 B T P1 )x(t) + (P2 M − P1 BR −1 B T P2 )z(t)

+ (P3 G + P1 DH − P1 BR −1 B T P3 )w(t) (3.12)
On the other hand, from the second equation of (3.8) and (3.9), we have
−λ̇(t) = (C T QC + AT P1 )x(t) + (AT P2 − C T QN )z(t) + AT P3 w(t) (3.13)

Then, from (3.12) and (3.13), we obtain
(P1 A + AT P1 − P1 BR −1 B T P1 + C T QC)x(t)
+ (AT P2 − C T QN + P2 M − P1 BR −1 B T P2 )z(t)
+ (AT P3 + P3 G + P1 DH − P1 BR −1 B T P3 )w(t) = 0 (3.14)
Notice the fact that (3.14) is true for any x(t), z(t), and w(t). Then, one yields
the Riccati equation as
P1 A + AT P1 − P1 BR −1 B T P1 + C T QC = 0 (3.15)
and the Sylvester equations
(AT − P1 BR −1 B T )P2 + P2 M = C T QN (3.16)
and
(AT − P1 BR −1 B T )P3 + P3 G = −P1 DH, (3.17)
respectively.
By Assumption 1, there exists a unique positive-definite solution P1 to the Riccati
equation (3.15).
It is clear from optimal control theory that A − BR −1 B T P1 is a Hurwitz matrix.
Then, by Lemma 2.3, Assumption 2 and the condition (2.20) guarantee that the
Sylvester equations (3.16) and (3.22) have unique solutions P2 and P3 , respectively.
Based on the above analysis, we present the following proposition which
provides the design method of the feedforward and feedback optimal tracking
controller.
Proposition 3.1 Consider the optimal tracking control problem of the offshore
platform system (2.3) with (3.1) subject to the quadratic performance index (3.4).
Under Assumptions 1 and 2, there exists a unique feedforward and feedback optimal
tracking control law of the form (3.10), where the matrix P1 is the unique positive-
definite solution to the Riccati matrix equation (3.15), and the matrices P2 and P3
are the unique solutions to the Sylvester equations (3.16) and (3.22), respectively.
Remark 3.1 The feedforward term −R −1 B T P3 w(t) in (3.10) is utilized to attenuate

the wave-induce vibration of the offshore platform. Specifically, if the feedforward
compensation term is ignored, then one yields a traditional optimal tracking control
law as
u(t) = −R −1 B T [P1 x(t) + P2 z(t)] (3.18)

3.3 Simulation Results and Discussions 37
Remark 3.2 Suppose that the quadratic performance index (3.4) is modified as
T
1
J = lim x T (t)Q̄x(t) + uT (t)R̄u(t) dt (3.19)
T →∞ 2T 0
where Q̄ is a 4 × 4 positive semi-definite matrix and R̄ > 0 is a scalar. Then, similar

to the design process of the feedforward and feedback optimal tracking control law
proposed above, a feedforward and feedback optimal control law of the system (2.3)
subject to quadratic performance index (3.19) can be obtained as
u(t) = −R̄ −1 B T [P̄1 x(t) + P̄3 w(t)] (3.20)
where P̄1 is the unique positive-definite solution to the Riccati equation
P̄1 A + AT P̄1 − P̄1 B R̄ −1 B T P̄1 + Q̄ = 0 (3.21)
and P̄3 satisfies the following Sylvester equation:
(AT − P̄1 B R̄ −1 B T )P̄3 + P̄3 G = −P̄1 DH (3.22)
Remark 3.3 In terms of implementation, if the state w(t) of the exogenous sys-
tem (2.18) is unavailable in the practical engineering, under Assumption 3, one
can design a disturbance-observer-based feedforward and feedback near optimal
tracking control law as
u(t) = −R −1 B T [P1 x(t) + P2 z(t) + P3 ŵ(t)] (3.23)
where ŵ(t) is the state estimation of the exogenous system (2.18) and satisfies

˙
ŵ(t) = Gŵ(t) + Θ f (t) − H ŵ(t) , ŵ(0) = ŵ0 (3.24)
with ŵ0 is the initial state of the disturbance observer (3.24), andΘ is the designed
observer matrix with appropriate dimensions.
Correspondingly, for the feedforward and feedback optimal control law (3.20),
one can design a disturbance-observer-based feedforward and feedback near optimal
control law as
u(t) = −R̄ −1 B T [P̄1 x(t) + P̄3 ŵ(t)] (3.25)
3.3 Simulation Results and Discussions
To show effectiveness of feedforward and feedback optimal tracking control

schemes for the offshore steel jacket platform, in this section, some simulation
results will be presented. A comparison of the optimal tracking control scheme
Table 3.1 Parameters of the offshore platform (2.3) and the wave force
Description Symbol Value Unit
Length of the offshore platform L 249 m
Diameter of the cylinder D̃ 1.83 m
Mass of the offshore platform m1 7,825,307 kg
Mass of the AMD m2 78,253 kg
Natural frequency of the offshore platform ω1 2.0466 rad/s
Natural frequency of the AMD ω2 2.0074 rad/s
Damping ration of the offshore platform ξ1 0.02 –
Damping ration of the AMD ξ2 0.2 –
Significant wave height Hs 7 m
Water depth d 218 m
Peak frequency of wave ω0 0.79 rad/s
Drag coefficient Cd 1.0 –
inertia coefficient Cm 1.5 –
Peakedness coefficient γ̄ 3.3 –
with and without feedforward compensation will be made first. Then, the feed-
forward and feedback optimal tracking control scheme will be compared with the
feedforward and feedback optimal control scheme [81] and the superiority of the
proposed optimal tracking control scheme with feedforward compensation will be
demonstrated.
In Fig. 2.1, the values of the masses, natural frequencies and the damping ratios
of the offshore platform with the AMD, and the wave-related parameters are from
[86], which are presented in Table 3.1. Based on these settings, the matrices A, B,
and D in (2.42) can be obtained as
⎧ ⎡ ⎤
⎪
⎪ 0 0 1.0000 0
⎪ ⎢
⎪
⎪
⎪ ⎢ 0 0 0 1.0000 ⎥
⎥
⎨A = ⎣
⎪
−4.2290 0.0403 −0.0899 0.0080 ⎦
−4.0297 0.8030 −0.8030 (3.26)
⎪
⎪ 4.0297
⎪
⎪ −4
T
⎪
⎪ B = 10 × 0 0 −0.0013 0.1278
⎪
⎩ T
D = 10−6 × 0 0 0.1278 0
T
The initial state x0 of the system is given as x0 = −0.02 0.03 0.01 0.03 . To
compute the wave force, let n = 7 in (2.14). Then based on the values given by
Table 3.1, one yields the matrices G̃ and H̃ in (2.19) as
G̃ = diag{ − 0.12571, −0.50283, −1.1314, −2.0113,

− 3.1427, −4.5254, −6.1596}, H̃ = −1.3803
6
x 10
8
2
Wave Force (N)
−2
−4
−6
−8
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.1 Wave force acting on the offshore platform
Furthermore, from (2.18), we can obtain the total wave force acting on the offshore
platform, which is shown in Fig. 3.1.
Due to the irregular nature of the wave force applied, both the peak values and the
root mean square (RMS) values of displacement, velocity of the offshore platform,
and the control force are investigated. Let Md , Mv , and Mu represent the peak
values of displacement, velocity of the offshore platform, and the required control
force, respectively, and Jd , Jv , and Ju denote the RMS values of displacement,
velocity of the offshore platform, and the control force, respectively, where
⎧
⎪
⎪ M = max{|x1 (t)|, t ∈ [0, T ]}, Jd = 1 T 2
⎪
⎨ d T 0 x1 (t)dt

1 T 2
⎪ Mv = max{|ẋ1 (t)|, t ∈ [0, T ]}, Jv = T 0 ẋ1 (t)dt
(3.27)
⎪
⎪
⎩ M = max{|u(t)|, t ∈ [0, T ]}, J = 1 T 2
u u T 0 u (t)dt
with T is a given measurement period.
3.3.1 Performance of System with Optimal Tracking Controller
In this subsection, a feedforward and feedback optimal tracking controller (FFOTC)

and a feedback optimal tracking controller (FOTC) will be designed to control the
offshore platform. The peak and root mean square (RMS) values of the oscillation
amplitudes of displacement and velocity and the required control force under no
control, FOTC, and FFOTC will be compared to demonstrate the superiority of the
proposed feedforward and feedback optimal tracking control scheme.
First, when no controller is applied to the offshore platform, it can be computed
that the peak values of the displacement and the velocity of the offshore platform
are 0.5295 m and 0.7327 m/s, respectively, and the RMS values of the displacement
and the velocity are 0.1921 m and 0.3108 m/s, respectively.
Then, we turn to design a feedforward and feedback optimal tracking controller
(FFOTC). For this, in the performance index (3.4), set
Q = 108 I2 , R = 10−9 (3.28)
where I2 is the 2 × 2 identity matrix.

In the exogenous system (3.3), the matrices M and N are given as
M = −0.2I2 , N = I2 (3.29)
Solving the Riccati equation (3.21), Sylvester equations (3.16) and (3.22) yield
⎧ ⎡ ⎤
⎪
⎪ 169750000 −72378 −20347000 −244140
⎪
⎪ ⎢ −72378
⎪
⎪ 259.84 39127 391.27 ⎥
⎪
⎪ P1 = ⎢
⎣ −20347000
⎥
⎪
⎪
⎪ 39127 11158000 92822 ⎦
⎨
−244140 391.27 92822 992.7
⎡ ⎤
⎪
⎪ −10105000 97979000
⎪
⎪
⎪
⎪ ⎢ −27000 −5000 ⎥
⎪
⎪ P2 = ⎢
⎣ −21262000
⎥ , P3 = Υ1 Υ2 Υ3
⎪
⎪
⎪ −4252000 ⎦
⎩
−211000 −42000
where
⎧ ⎡ ⎤
⎪
⎪ 757520 373220 128710 65080 40740
⎪
⎪ ⎢ −680
⎪
⎪
⎪ ⎢ −840 −440 −250 −160 ⎥ ⎥
⎪
⎪ Υ 1 = ⎣ −243510 −171050 −69580 −35900 −22280 ⎦
⎪
⎪
⎪
⎪
⎪
⎪ −2630 −1810 −730 −370 −230
⎪
⎪ ⎡ ⎤
⎪
⎪ 28190 20610 756970 791380 389170
⎪
⎨ ⎢ −110 −80 700 −560 −500 ⎥
Υ2 = ⎢ ⎥
⎣ −15410 −11430 −88820 −242420 −140740 ⎦
⎪
⎪
⎪
⎪
⎪
⎪ −150 −110 −1060 −2630 −1520
⎪
⎪ ⎡ ⎤
⎪
⎪ 216010 137380 95310 70090
⎪
⎪
⎪
⎪ ⎢ −320 −220 −150 −110 ⎥
⎪
⎪ Υ3 = ⎢ ⎥
⎪
⎪ ⎣ −81180 −52090 −36180 −26580 ⎦
⎪
⎪
⎩
−870 −560 −390 −290
0.6
No control
FOTC
FFOTC
0.4
Displacement of the Offshore Platform (m)
0.2
−0.2
−0.4
−0.6
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.2 Displacement of the offshore platform under no control, FOTC, and FFOTC
1
No control
0.8 FOTC
FFOTC
0.6
Velocity of the Offshore Platform (m/s)
0.4
0.2
−0.2
−0.4
−0.6
−0.8
−1
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.3 Velocity of the offshore platform under no control, FOTC, and FFOTC
Then an FFOTC (3.10) and a feedback optimal tracking controller (FOTC) with
the form (3.18) are obtained, respectively. Under no control, FFOTC, and FOTC,
the response curves of the displacement, the velocity of the system (2.3), and
the required control force are presented in Figs. 3.2, 3.3, and 3.4, respectively.
The tracking errors of the displacement and the velocity of the offshore platform
without control and under FFOTC and FOTC are depicted in Figs. 3.5 and 3.6,
7
x 10
5
FOTC
4 FFOTC
2
Control Force (N)
−1
−2
−3
−4
−5
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.4 Control force required by FOTC and FFOTC
1
FOTC
FFOTC
0.8
0.6
Tracking Error of the Displacement
0.4
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.5 Tracking error of the displacement under FOTC and FFOTC
respectively. From these two figures, one can see that FFOTC has a smaller tracking
error than FOTC. It also can be seen from Figs. 3.2, 3.3, and 3.4 that FFOTC can
significantly reduce the oscillation amplitudes of the displacement and the velocity
of the offshore steel jacket platform. In addition, the required control by FFOTC is
much smaller than the one by FOTC.
0.8
FOTC
FFOTC
0.6
0.4
Tracking Error of the Velocity
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.6 Tracking error of the velocity under FOTC and FFOTC
In fact, it can be computed that under FOTC, the peak values of the displacement
and the velocity of the offshore platform are 0.2104 m and 0.2003 m/s, respectively,
and the maximum control force is about 4.2738 ×107 N. The RMS values of the
displacement and the velocity of the offshore platform are 0.1137 m and 0.0901 m/s,
respectively, and the RMS value of the control force is 2.4552 ×107 N. Under
FFOTC, the peak values of the displacement, the velocity of the offshore platform,
and the required control force are 0.0298 m, 0.0214 m/s, and 2.6898 ×107 N,
respectively, and the RMS values of the displacement, the velocity of the offshore
platform, and the control force are 0.0211 m, 0.0171 m/s, and 1.6241 ×107 N,
respectively.
To compare the two different control schemes clearly, the peak and RMS values
of the offshore platform and the required control force without control and under
FOTC and FFOTC are listed in Table 3.2. From this table, one can obtain the
following facts.
• Under FOTC, the peak values of the oscillation amplitudes of the displacement
and the velocity of the offshore platform are reduced by 60% and 72%,
respectively, while under FFOTC, there are further reduced by 94% and 97%,
respectively.
• From the point of view of the RMS values, it can be obtained that under FOTC,
the RMS values of the displacement and the velocity are reduced by 41% and
71%, respectively, while under FFOTC, there are even reduced by 89% and 94%,
respectively.
Table 3.2 Peak and RMS values of displacement, velocity of the offshore platform, and the
required control force without control and under FOTC and FFOTC
Peak value RMS value
Controller
Md (m) Mv (m/s) Mu (107 N) Jd (m) Jv (m/s) Ju (107 N)
No control 0.5295 0.7327 – 0.1921 0.3108 –
FOTC 0.2104 0.2003 4.2738 0.1137 0.0901 2.4552
FFOTC 0.0298 0.0214 2.6898 0.0211 0.0171 1.6241
• Moreover, both the peak and RMS values of the control force required by FOTC
are 1.5 times as the ones by FFOTC. It indicated that FFOTC is more efficient
than FOTC to attenuate the wave-induced vibration of the offshore platform.
3.3.2 Comparison of Optimal Controller and Tracking

Controller
To compare the optimal tracking control scheme proposed in this paper with the
optimal control scheme in [81], based on the design process in [81], we first
design a feedforward and feedback optimal controller (FFOC). For this, discretizing
system (2.3) with a sampling period 0.1 yields
x(k + 1) = Ãx(k) + B̃u(k) + D̃f (k), k = 0, 1, 2, · · · , x(0) = x0 (3.30)
where
⎧ ⎡ ⎤
⎪
⎪ 0.9790 0.0002 0.09885 0.00005
⎪
⎪ ⎢ 0.0189 0.9805 0.00453 0.09545 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ Ã = ⎣ −0.4179 0.0038 0.9701 0.00095 ⎦
⎪
⎪
⎪
⎨
0.3655 −0.3844 0.09522 0.90380
⎡ ⎤ ⎡ ⎤
⎪
⎪ −6.16 × 10−10 6.348 × 10−10
⎪
⎪
⎪ ⎢ 6.199 × 10−8 ⎥ ⎢ −11 ⎥
⎪
⎪
⎪ B̃ = ⎢ ⎥ , D̃ = ⎢ 1.876 × 10 ⎥
⎪
⎪ ⎣ −1.205 × 10−8 ⎦ ⎣ 1.263 × 10−8 ⎦
⎪
⎩
1.219 × 10−6 5.783 × 10−10
Discretizing the exogenous system (2.18), we obtain
w(k + 1) = Ĝw(k), f (k) = H w(k), k = 0, 1, 2, · · · , (3.31)
where

Ĝ11 Ĝ12
Ĝ = (3.32)
Ĝ21 Ĝ11
with
⎧
⎨ Ĝ11 = diag{1.0000, 1.0000, 0.9999, 0.9999, 0.9998, 0.9998, 0.9997}
Ĝ = diag{0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01}
⎩ 12
Ĝ21 = −diag{0.0013, 0.0050, 0.0113, 0.0201, 0.0314, 0.0453, 0.0616}
(3.33)
In the following discrete quadratic performance index [81]
1 T
N
Jˆ = lim [x (k)Q̂x(k) + u(k)R̂u(k)]dt (3.34)
N →∞ N
k=0
set the weight matrices Q̂ and R̂ as
Q̂ = diag{5 × 107 , 1, 4 × 109 , 1}, R̂ = 9.5 × 10−11 (3.35)
Then a FFOC is obtained as
u(k) = K̂1 x(k) + K̂2 f (k) + +K̂3 w(k) (3.36)
where
⎧
⎪
⎨ K̂1 = 10 × 1.4675 0.0311 7.9532 0.0045 , K̂2 = 1.0042
7
K̂3 = 105 × 2.0250 2.0430 2.0698 2.0890 2.0601 1.9321 1.6938

⎪
⎩
0.3358 0.3505 0.3799 0.4273 0.4871 0.5385 0.5548
Applying the FFOC to the offshore platform, it can be computed that the peak
values of the oscillation amplitudes of the displacement and the velocity of the
offshore platform are 0.2189 m and 0.0244 m/s, respectively, and the RMS values
of the displacement and the velocity are 0.1037 m and 0.0112 m/s, respectively.
The peak and RMS values of the control force are about 4.3327 ×107 N and
2.0153 ×107 N, respectively, which are listed in Table 3.3. The responses of the
displacement and the velocity of the offshore platform and the required control
force by FFOC and FFOTC are depicted in Figs. 3.7, 3.8, and 3.9, respectively.
From Table 3.3 and the figures, one can see that:
• The oscillation amplitudes of the velocity of the offshore platform under FFOC
and FFOTC are almost in the same level, while the oscillation amplitudes of the
displacement of the offshore platform under FFOTC are much smaller than the
one under FFOC.
• Moreover, the control force required by FFOTC is smaller than that by FFOC.
To sum up, it can be clearly observed that the proposed optimal tracking con-
troller with feedforward compensation can significantly reduce the wave-induced
vibration of the offshore platform and thereby improve the control performance of
Table 3.3 Peak and RMS values of displacement, velocity of the offshore platform and the
required control force under FFOC [81] and FFOTC
Controller
Md (m) Mv (m/s) Mu (107 N) Jd (m) Jv (m/s) Ju (107 N)
FFOC 0.2189 0.0244 4.3327 0.1037 0.0112 2.0153
FFOTC 0.0298 0.0214 2.6898 0.0211 0.0171 1.6241
0.3
FFOC
0.25 FFOTC
0.2
Displacement of the Offshore Platform (m)
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
−0.25
−0.3
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.7 Displacement of the offshore platform under FFOC and FFOTC
0.05
FFOC
0.04 FFOTC
Velocity of the the Offshore Platform (m/s)
0.03
0.02
0.01
−0.01
−0.02
−0.03
−0.04
−0.05
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.8 Velocity of the offshore platform under FFOC and FFOTC
3.5 Notes 47
8
x 10
1
FFOC
0.8 FFOTC
0.6
0.4
Control Force (N)
0.2
−0.2
−0.4
−0.6
−0.8
−1
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.9 Control force required by FFOC and FFOTC
the offshore platform. Furthermore, the designed FFOTC is better than the FOTC
and FFOC [81] in two respects: vibration amplitudes of the displacement and the
velocity of the offshore platform and the required control force.
3.4 Conclusions
This chapter presents a feedforward and feedback optimal tracking control scheme
for the offshore platform. The gain matrices of the optimal tracking controller
can be obtained by solving an algebraic Riccati equation and Sylverter equations,
respectively. Simulation results have shown that under the feedforward and feedback
optimal tracking controllers, both wave-induced vibration amplitudes of the offshore
steel jacket platform and the required control force are significantly reduced.
3.5 Notes
Optimal control schemes have been extensively utilised to the active control of
offshore platforms. The earlier results concerning optimal control for offshore
platforms can be found in Kawano [71], Yoshida et al. [76], Abdel-Rohman [11],
Suneja et al. [75], Ahmad et al. [77], Terro [72], and Alves et al. [78]. The optimal
controllers in the frequency domain are designed by Yoshida et al. [76] and Mahadik
et al. [73]. A nonlinear stochastic optimal control scheme for a jacket platform
is developed by Luo et al. [74]. By using feedforward compensation and optimal
control strategy, feedfoward and feedback optimal controllers are advocated by
Wang et al. [80], Ma et al. [81, 82], and Zhang et al. [85], where the key point
of controller design is that the wave force acting on offshore platforms is modeled
as an output of a linear exogenous system.
Based on [83], this chapter presents a feedforward and feedback optimal tracking
control scheme for a jacket-type offshore platform subject to wave force. Note that
discrete control schemes are easy to implement for computer control systems. For
the offshore steel jacket platform with control delays, a discrete feedforward and
feedback optimal tracking control scheme with memory is developed by Zhang et
al. [84]. As for the detailed design process and analysis of the control scheme, we
refer to Zhang et al. [84].
Indeed, optimal control is effective to reduce vibration of offshore platforms
and thereby can improve the performance of the offshore platform systems. How-
ever, optimal control generally requires the exact dynamic model of the offshore
platforms. Consequently, the practical implementation of controller and the control
effects are confined. It is still challenging to develop effective optimal controllers to
uncertain and general dynamic models of offshore platforms, which is an important
research topic in the future.
Note that offshore platforms are inevitably affected by ocean waves, ice, winds,
flow, and even earthquakes. The external loading generally results in random
features of dynamics of offshore platforms. Applying stochastic control theory to
active control of offshore platform is a natural and feasible way [74]. However,
few results about stochastic control for offshore platforms are by far available in
the literature. Therefore, for the offshore platform systems, some issues including
system modeling, filtering, and controller design in the stochastic control theory
framework deserve further exploration.
Chapter 4
Integral Sliding Mode H∞ Control
In this chapter, sliding mode H∞ control for an offshore steel jacket platform subject
to nonlinear self-excited wave force and external disturbance is developed. A sliding
mode H∞ controller is designed to reduce oscillation amplitudes of the offshore
platform. In the case that the dynamic model of the offshore platform is subject to
parameter perturbations, a robust sliding mode H∞ control scheme is proposed. It is
found through simulation results that compared with an H∞ controller and a sliding
mode controller, the sliding mode H∞ controller requires much less control force;
and the oscillation amplitudes of the offshore platform under the sliding mode H∞
controller are less than those under the sliding mode controller.
The rest of this chapter is organized as follows. In Sect. 4.1, the problem of
sliding mode H∞ controller design for an offshore platform system subject to
external disturbance is formulated. Section 4.2 presents the sliding surface design
and stability analysis of the resulting sliding motion. In Sects. 4.3 and 4.4, the design
results of the sliding mode H∞ controller and the robust sliding mode H∞ controller
are presented, respectively. The simulation results are given in Sect. 4.4 to illustrate
the usefulness and the advantages of the proposed methodology, conclusions are
given in Sect. 4.6, and a note is given in Sect. 4.7.
4.1 Problem Formulation
Consider the offshore platform system (2.50), where the external disturbance
on the offshore platform are considered. To design a controller to reduce the
oscillation amplitudes of the system, one should specify the control output so that
the performance index from the external disturbance to the control output can be
realized with the specified requirement. For this aim, the control output is given as
η(t) = C1 x(t) + D1 ζ (t), (4.1)

https://doi.org/10.1007/978-981-13-2986-9_4
50 4 Integral Sliding Mode H∞ Control
where C1 and D1 are problem-dependent constant matrices with appropriate

dimensions. For instance, if we want the first two displacement responses to be
controlled, we can choose

100000 0.1 0
C1 = , D1 =
001000 0 0.1
In what follows, a sliding mode H∞ control scheme is developed such that the
system (2.50) with (4.1) satisfies
(i) in the designed sliding surface, the resulting closed-loop system is asymptoti-
cally stable; and under the zero initial condition, the H∞ performance
η(t) < γ ζ (t) (4.2)
of the closed-loop system is guaranteed for nonzero ζ (t) ∈ L2 [0, ∞) and a

prescribed γ > 0; and
(ii) under the designed sliding mode H∞ control law, the state trajectory of
system (2.50) can be driven into the sliding surface in finite time and maintain
on it thereafter.
To obtain the main results, the following assumption is needed.
Assumption 1 The matrix D1 in (4.1) and the H∞ performance level γ are
assumed to satisfy the constraint as
D1T D1 < γ 2 I (4.3)
4.2 Sliding Surface Design and Stability Analysis of Sliding

Motion
The sliding surface is designed as

t
s(t) = Gx(t) − G(A + BK) x(θ )dθ (4.4)
0
where G is 1 × 6 real matrix to be chosen such that GB is non-singular; K is 1 × 6

gain matrix to be determined. If the matrix GB is non-singular, one obtains the
equivalent control ueq (t) as
ueq (t) = Kx(t) − (GB)−1 G[Df (x(t), t) + D0 ζ (t)] (4.5)
which leads to the sliding motion

4.2 Sliding Surface Design and Stability Analysis of Sliding Motion 51
ẋ(t) = (A + BK)x(t) + D̄f (x(t), t) + D̄0 ζ (t) (4.6)
where D̄ = ḠD and D̄0 = ḠD0 with Ḡ = [I − B(GB)−1 G].

Proposition 4.1 Under Assumption 1, for given scalars μ > 0 and γ > 0,
the sliding motion (4.6) with ζ (t) = 0 is asymptotically stable, and the H∞
performance (4.2) is guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and a prescribed
γ > 0 if there exist a 6 × 6 real matrix P > 0 and a 1 × 6 real matrix K such that
⎡ ⎤
Λ P D̄ P D̄0 μI C1T
⎢ ∗ −I ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I 0 D1T ⎥<0 (4.7)
⎢ ⎥
⎣∗ ∗ ∗ −I 0 ⎦
∗ ∗ ∗ ∗ −I
where Λ = P A + AT P + P BK + K T B T P .
Proof We first consider the asymptotic stability of the system (4.6) with ζ (t) = 0
described by
ẋ(t) = (A + BK)x(t) + D̄f (x(t), t) (4.8)
Choose a Lyapunov candidate as
V1 (x(t)) = x T (t)P x(t) (4.9)
where P > 0. Taking the derivative of V1 (x(t)) with respect to t along the
trajectory of system(4.8), noting that (2.43), and introducing a new vector α T (t) =
T T
x (t) f (x(t), t) , one obtains
V̇1 (x(t)) ≤ α T (t)Ψ α(t) (4.10)
where

Λ + μ2 I P D̄
Ψ = (4.11)
∗ −I
If the matrix inequality (4.7) holds, then by Schur complement, we have Ψ < 0,
which means that the system (4.8) is asymptotically stable.
In the following, we prove that the H∞ performance is guaranteed for nonzero
ζ (t) ∈ L2 [0, ∞] under zero initial condition.
Taking the derivative of V1 (x(t)) with respect to t along the trajectory of sliding
motion (4.6), and noticing that (2.43), one yields
V̇1 (x(t)) ≤x T (t)(Λ + μ2 I )x(t) − f T (x(t), t)f (x(t), t)

+ 2x T (t)P D̄f (x(t), t) + 2x T (t)P D̄0 ζ (t) (4.12)

Letting β T (t) = x T (t) f T (x(t), t) ζ T (t) and combining with (4.1), after simple
manipulation, we obtain
V̇1 (x(t)) + ηT (t)η(t) − γ 2 ζ T (t)ζ (t) ≤ β T (t)Ξβ(t) (4.13)
where
⎡ ⎤
Λ + μ2 I + C1T C1 P D̄ P D̄0 + C1T D1
Ξ =⎣ ∗ −I 0 ⎦ (4.14)
∗ 0 D1 D1 − γ I
T 2
It is clear that if the matrix inequality (4.7) is feasible, then applying the Schur
complement yields Ξ < 0, which leads to
V̇1 (x(t)) + ηT (t)η(t) − γ 2 ζ T (t)ζ (t) < 0 (4.15)
Integrating both sides of (4.15) from 0 to ∞, noting the fact that V1 (x(0)) = 0 under
zero initial condition, we have
∞
[ηT (t)η(t) − γ 2 ζ T (t)ζ (t)]dt < 0 (4.16)
0
which means the H∞ performance (4.2) is guaranteed. This completes the proof.

4.3 Design of the Sliding Mode H∞ Control Law
In Proposition 4.1, a sufficient condition of the asymptotic stability and the

prescribed disturbance attenuation level γ is obtained for sliding motion (4.6). We
now proceed to synthesize a sliding mode H∞ control law such that the reachability
of the specified sliding surface is ensured.
The sliding mode H∞ control law is designed as
s(t)
u(t) = Kx(t) − (GB)−1 ρ(x(t), t) (4.17)
|s(t)|
where | · | represents the absolute value and ρ(x(t), t) is a function satisfying
ρ(x(t), t) = μGDx(t) + GD0 ζ (t) + δ (4.18)
with δ > 0 is a design parameter.

4.3 Design of the Sliding Mode H∞ Control Law 53
Proposition 4.2 Consider the system (2.50) with (4.1). If the sliding surface is
given by (4.4), where K is the solution of matrix inequality (4.7), then the
reachability of sliding surface s(t) = 0 is ensured by the sliding mode H∞ control
law (4.17).
Proof To analyze the reachability, we choose the Lyapunov function candidate as
1 2
V2 (s(t)) = s (t) (4.19)
2
From (2.50), (4.4), and (4.17), it follows that
s(t)
ṡ(t) = GDf (x(t), t) + GD0 ζ (t) − ρ(x(t), t) (4.20)
|s(t)|
Thus, from (4.19) and (4.20), together with (4.18), we have
V̇2 (s(t)) ≤ −δ|s(t)| (4.21)
which implies that under the control law (4.17), the trajectories of the system (2.50)
will be driven onto the specified sliding surface s(t) = 0.

Note that Proposition 4.1 provides a stability condition for sliding motion (4.6).
It is clear that the condition is nonlinear due to the nonlinear term P BK. In
order to obtain the controller gain K in (4.17), pre- and post-multiplying (4.7) by
diag{P −1 , I, I, I, I }, respectively, and setting P̄ = P −1 and K̄ = KP −1 , we
have the following equivalent version of Proposition 4.1.
Proposition 4.3 Under Assumption 1, for given scalars μ > 0 and γ > 0, the
sliding motion (4.6) is asymptotically stable, and the H∞ performance (4.2) is
guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and the prescribed γ > 0 if there exist a
6 × 6 real matrix P̄ > 0 and a 1 × 6 real matrix K̄ such that
⎡ ⎤
Λ̄ D̄ D̄0 μP̄ P̄ C1T
⎢ ∗ −I 0 0 ⎥
⎢ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I 0 D1T ⎥ < 0 (4.22)
⎢ ⎥
⎣∗ ∗ ∗ −I 0 ⎦
∗ ∗ ∗ ∗ −I
where
Λ̄ = AP̄ + P̄ AT + B K̄ + K̄ T B T (4.23)
Moreover, the gain matrix K of the control law (4.17) is given by K = K̄ P̄ −1 .

As two special cases of the sliding mode H∞ control law, based on Proposi-
tion 4.3, a sliding mode control law and an H∞ control law can be easily designed
by following Corollaries.
Corollary 4.1 For a given scalar μ > 0, if there exist a 6 × 6 real matrix P̄ > 0
and a 1 × 6 real matrix K̄ such that
⎡ ⎤
Λ̄ D̄ μP̄
⎣ ∗ −I 0 ⎦ < 0 (4.24)
∗ ∗ −I
then, under the sliding mode control law (4.17), the system (2.50) is asymptotically
stable in the sliding surface s(t) = 0; and the reachability of the sliding surface is
guaranteed. Moreover, the gain K of the control law is given by K = K̄ P̄ −1 .
Corollary 4.2 Under Assumption 1, for given scalars μ > 0 and γ > 0, if there
exist a 6 × 6 real matrix P̄ > 0 and a 1 × 6 real matrix K̄ such that
⎡ ⎤
Λ̄ D D0 μP̄ P̄ C1T
⎢ ∗ −I 0 0 ⎥
⎢ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ I 0 D1T ⎥ < 0
2 (4.25)
⎢ ⎥
⎣∗ ∗ ∗ −I 0 ⎦
∗ ∗ ∗ ∗ −I
then, under the H∞ control law
u(t) = Kx(t) (4.26)
the system (2.50) with (4.1) is asymptotically stable; and the H∞ performance (4.2)
is guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and the prescribed γ > 0. Moreover,
the gain K of the control law (4.26) is given by K = K̄ P̄ −1 .
4.4 Design of the Robust Sliding Mode H∞ Control Law
In this subsection, we intend to design a robust sliding mode H∞ control law for an
uncertain system (2.47).
Suppose that the sliding surface function is designed as the same form as (4.4).
Then, an equivalent control can be written as
ueq (t) = Kx(t) − (GB)−1 G[M̂ F̂ (t)N̂x(t) + Df (x(t), t) + D0 ζ (t)] (4.27)

4.4 Design of the Robust Sliding Mode H∞ Control Law 55
Thus, the resulting sliding motion can be written as
ẋ(t) = [A + BK + M̄1 Δ1 (t)N1 ]x(t) + D̄f (x(t), t) + D̄0 ζ (t) (4.28)
where M̂¯ = ḠM̂.

The sliding mode H∞ control law is designed in the form
s(t)
u(t) = Kx(t) − (GB)−1 ρ1 (x(t), t) (4.29)
|s(t)|
where
ρ1 (x(t), t) = [μGD + GM̂N̂ ]x(t) + GD0 ζ (t) + δ (4.30)
The following proposition provides a feasibility of designing a robust sliding

mode H∞ control law for uncertain system (2.47).
Proposition 4.4 Under Assumption 1, for given scalars μ > 0 and γ > 0, if there
exist a 6 × 6 real matrix P̄ > 0, a 1 × 6 real matrix K̄, and a scalar ε > 0 such that
⎡ ⎤
Λ̄ D̄ D̄0 μP̄ P̄ C1T εM̂¯ P̄ N̂ T
⎢ ∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I 0 D1T 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ −I 0 0 0 ⎥<0 (4.31)
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
where Λ̄ is given by (4.23). Then, the sliding motion (4.28) is robustly stable, and the
H∞ performance (4.2) is guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and a prescribed
γ > 0; the reachability of the sliding surface s(t) = 0 is guaranteed by the sliding
mode H∞ control law (4.29). Moreover, the gain matrix K of the control law (4.29)
is given by K = K̄ P̄ −1 .
Proof The proof are similar to those in Propositions 4.1 and 4.2 and thus are omitted
here.
Correspondingly, the corollaries stated below provide approaches to designing
robust sliding mode control law and robust H∞ control law, respectively.
Corollary 4.3 For a given scalar μ > 0, if there exist a 6 × 6 real matrix P̄ > 0, a
1 × 6 real matrix K̄, and a scalar ε > 0 such that
⎡ ⎤
Λ̄ D̄ μP̄ εM̂¯ P̄ N̂ T
⎢ ⎥
⎢∗ −I 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ −I 0 0 ⎥ < 0, (4.32)
⎢ ⎥
⎣∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ −εI
then, under the robust sliding mode control law (4.29), the uncertain system (2.47)
is robustly stable in sliding surface s(t) = 0; and the reachability of sliding surface
is guaranteed; and the gain matrix K of the control law is given by K = K̄ P̄ −1 .
Corollary 4.4 Under Assumption 1, for given scalars μ > 0 and γ > 0, if there
exist a 6 × 6 real matrix P̄ > 0, a 1 × 6 real matrix K̄, and a scalar ε > 0 such that
⎡ ⎤
Λ̄ D D0 μP̄ P̄ C1T M̂¯ P̄ N̂ T
⎢ ∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I 0 D1T 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ −I 0 0 0 ⎥ < 0, (4.33)
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
then, under the robust H∞ control law in the form as (4.26), the uncertain
system (2.47) with (4.1) is robustly stable; and the H∞ performance (4.2) is
guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and the prescribed γ > 0. Moreover,
the gain matrix K of the control law is given by K = K̄ P̄ −1 .
4.5 Simulation Results
In this section, we will first design a sliding mode H∞ controller (SMHC) formed
as (4.17) for the system (2.50) with (4.1) to show effectiveness of the proposed slid-
ing mode H∞ control scheme. The performances of the system under the SMHC,
H∞ controller (HIC) and sliding mode controller (SMC) are compared. Then,
for the uncertain system (2.47) with (4.1), a robust sliding mode H∞ controller
(RSMHC) defined in (4.29) will be given to improve the control performance.
4.5.1 System Parameters of an Offshore Platform
An offshore platform with a TMD mechanism presented in Fig. 2.2 is simulated. The
parameters of the system and the waves are given in Table 4.1, which are derived
from [65] and [72]. With the setting in the table, the nonlinear wave force f (x, t) can
4.5 Simulation Results 57
Table 4.1 Parameters of the offshore platform and the wave

Description Symbol Value Unit
Damping ration of mode 1 ξ1 0.005 –
Damping ration of mode 2 ξ2 0.005 –
Mode shape vector of mode 1 φ1 −0.003445 –
Mode shape vector of mode 2 φ2 0.00344628 –
Natural frequency of mode 1 ω1 1.8180 rad/s
Natural frequency of mode 2 ω2 10.8683 rad/s
Damping ration of TMD ξT 0.15 –
Natural frequency of TMD ωT 1.8180 rad/s
Stiffness of TMD KT 1551.5 N/m
Damping coefficient of TMD CT 256 Ns/m
Mass of TMD mT 469.4836 kg
Height of wave Hw 12.19 m
Length of wave λw 182.88 m
Frequency of wave ωw 1.8 rad/s
Depth of water hw 76.2 m
Density of water ρw 1025.6 kg/m3
be computed as Appendix A of [65], and system matrices A and B can be obtained

as
⎧ ⎡ ⎤
⎪
⎪ 0 1 0 0 0 0
⎪ ⎢ −3.3253 −0.0212 0.0184
⎪
⎪
⎪
⎪ ⎢ 0.0030 −5.3449 −0.8819 ⎥ ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 0 0 1 0 0 ⎥
⎪
⎪ A = ⎢ ⎥
⎪
⎪ ⎢ 0.0184 0.0030 −118.1385 −0.1118 5.3465 0.8822 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎣ 0 0 0 0 0 1 ⎦
⎪
⎨
⎡ −0.0114 −0.0019
⎤ 0.0114 0.0019 −3.3051 −0.5454
⎪
⎪ 0
⎪
⎪ ⎢ 0.003445 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 ⎥
⎪
⎪ B = ⎢ ⎥
⎪
⎪ ⎢ −0.00344628 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎣ ⎦
⎪
⎪ 0
⎩
0.00213
(4.34)
The nonlinear self-excited wave force f (x(t), t) can be computed as [65]. The
external disturbance force acting on the first mode is approximated by a uniformly
distributed random signal ranging between −4.6 × 105 N and 4.6 × 105 N, while
the external disturbance force acting on the second mode is approximated by a
uniformly distributed random signal ranging between −1.1×105 N and 1.1×105 N.
4.5.2 Performance of the Nominal System
In this subsection, under the SMC, HIC, and SMHC, the performance of the
system (2.50) with (4.1) will be analyzed.
Firstly, when no controller is applied to the system, the responses of the three
floors of the system are presented in Fig. 4.1. The oscillation amplitudes of the
first, second, and third floors peak to peak are 1.4159, 1.5270, and 1.6061 m,
respectively. The average response of the three floors peak to peak is 1.5164 m.
It can be found that the offshore platform is very dangerous to work. Secondly,
we consider the sliding mode control scheme. Set μ = 0.8, δ = 0.1 and
G = [500 1 1000 1 0 100]. By Corollary 4.1, we obtain the gain of controller
SMC as
KSMC = 105 × [−0.3312 0.0623 1.1104 − 0.0043 − 0.1709 − 0.3191].
When this SMC is applied to the system, the responses of the first, second, and third
floors and the required control force are shown in Fig. 4.2, from which one can see
that under the SMC, the oscillation amplitudes of the first, second, and third floors
peak to peak are 0.2333, 0.2537, and 0.2688 m, respectively. The control force peak
to peak by the SMC is about 1.8401 × 105 N.
Under the SMC, the variation of the sliding surface s(t) is illustrated in Fig. 4.3.
It can be found that the sliding function s(t) = 0 and the average value of the sliding
function s(t) are equal to zero. Theoretically, the sliding function should be zero,
Response of floor 1 (m)
1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
0 50 100 150 0 50 100 150
Time (s) Time (s)
0.5
-0.5
-1
0 50 100 150
Time (s)
Fig. 4.1 Responses of the three floors of the system (2.50) without control

0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 5
0.2 1.5
Control force (N)

0.1 1
0.5
0
0
-0.1 -0.5
-0.2 -1
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.2 Responses of the three floors of the nominal system (2.50) and the control force by SMC
500
400
300
Variation of sliding surface
200
100
-100
-200
-300
-400
-500
0 50 100 150
Time (s)
Fig. 4.3 Variation of the sliding surface under the SMC

s(t) = 0. However, due to the self-excited nonlinear wave force and the external
disturbance, there always exists the deviations in the sliding function.
Thirdly, we study the H∞ control scheme. Set γ = 0.2. By Corollary 4.2, we
obtain the gain of an HIC as
KHIC = 105 × [−0.7364 0.1291 2.0535 − 0.0555 − 0.2821 − 0.6061].
Under the HIC, the peak-to-peak oscillation amplitudes of the first, second, and third
floors are 0.2040, 0.2220, and 0.2358 m, respectively. The range of the required
control force by the HIC is about 2.6952 × 105 N. The responses of the three floors
and the required control force are presented in Fig. 4.4.
Finally, we turn to the sliding mode H∞ control scheme. Let γ = 0.2. By
Proposition 4.3, the gain of an SMHC is obtained as
KSMHC = 104 × [−1.1815 0.1230 3.0622 0.0836 − 0.3016 − 0.9181].
When the SMHC is applied to the system (2.50) with (4.1), the displacements
of the three floors and the required control force are presented in Fig. 4.5, and the
curve of the sliding function s(t) is given by Fig. 4.6. It can be seen that the peak-to-
peak oscillation amplitudes of the first, second, and third floors are 0.1998, 0.2177,
and 0.2317 m, respectively. The control force peak to peak by the SMHC is about
1.5449 × 105 N.
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 5
0.2 2
Control force (N)
0.1 1
0 0
-0.1 -1
-0.2 -2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.4 Responses of the three floors of the nominal system (2.50) and the control force by HIC

0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 4
0.2 10
Control force (N)

0.1
5
0
0
-0.1
-0.2 -5
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.5 Responses of the three floors of the nominal system (2.50) and the control force by SMHC
400
300
200
100
-100
-200
-300
-400
-500
0 50 100 150
Time (s)
Fig. 4.6 Variation of the sliding surface under the SMHC

Table 4.2 Maximum control forces and oscillation amplitudes of three floors of the system (2.50)
under different controllers
γ Controller Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force(105 N)
– No control 1.4159 1.5270 1.6061 –
– SMC 0.2333 0.2537 0.2688 1.8401
0.2 HIC 0.2040 0.2220 0.2358 2.6952
SMHC 0.1998 0.2177 0.2317 1.5449
0.3 HIC 0.2046 0.2226 0.2364 2.7102
SMHC 0.2027 0.2208 0.2352 1.5564
0.4 HIC 0.2036 0.2216 0.2354 2.0259
SMHC 0.2026 0.2204 0.2343 1.5137
0.5 HIC 0.2041 0.2221 0.2357 2.0242
SMHC 0.2043 0.2226 0.2371 1.6039
0.7 HIC 0.2040 0.2220 0.2357 2.0458
SMHC 0.2041 0.2221 0.2361 1.5480
In Table 4.2, in other cases of γ = 0.3, 0.4, 0.5 and 0.7, the oscillation
amplitudes of the system and the control force under the HIC and SMHC are
compared, where the performances of the system under the SMC and the case of
without control are also presented.
From Table 4.2 and Figs. 4.1, 4.2, 4.3, 4.4, and 4.5, one can see clearly that:
• The SMHC can reduce the oscillation amplitudes of the system to about 14%
of the oscillation amplitudes of the system without control while the SMC can
reduce the oscillation amplitudes of the system to about 17% of the oscillation
amplitudes of the system without control;
• Under the SMHC and the SMC, the average oscillation amplitude of the three
floors is about 0.22 and 0.25 m, respectively. Clearly, the oscillation amplitudes
of the three floors under the SMHC are less than those under the SMC. Moreover,
it can be found that the control force required by the SMHC is less than the one
by the SMC;
• Under the SMHC and the HIC, the average oscillation amplitude of the three
floors is about 0.22 m. However, it is not difficult to observe that the SMHC
requires less control force than the HIC.
4.5.3 Performance of the Uncertain System
In this subsection, the performance of the uncertain system (2.47) with (4.1) is
investigated when no controller, and a robust sliding mode controller (RSMC), a
robust H∞ controller (RHIC), and a robust sliding mode H∞ controller (RSMHC)
are applied, respectively.
Suppose that the maximum perturbation bounds ω̂i = 0.02, i = 1, 2, T , ξ̂1 =

ξ̂2 = 0.005 and ξ̂T = 0.01. Then by (2.48), the matrices M̂ and N̂ in system (2.47)
can be obtained as
⎧ ⎡ ⎤
⎪
⎪ 0 0 0 0 0 0
⎪ ⎢ 0.02 0.005 0
⎪
⎪
⎪
⎪ ⎢ 0 0 0 ⎥⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 0 0 0 0 0 ⎥
⎪
⎪ M̂ = ⎢ ⎥
⎪
⎪ ⎢ 0 0 0.02 0.005 0 0 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎣ 0 0 0 0 0 0 ⎦
⎪
⎨
⎡ 0 0 0 0 0.02 0.01 ⎤
⎪
⎪ −6.6102 −0.0182 0 0 0 0
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 −3.6360 0 0 0 0 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 0 −236.2399 −0.1087 0 0 ⎥
⎪
⎪ N̂ = ⎢ ⎥
⎪
⎪ ⎢ −21.7366 ⎥
⎪
⎪ ⎢
0 3.6360 0 0 0
⎥
⎪
⎪ ⎣ −0.0125 −0.0010 0.0125 0.0010 −6.6102 −0.5454 ⎦
⎪
⎪
⎩
0 −0.0125 0 0.0125 0 −3.6360
(4.35)
Let F̂ (t)=sin(t), and other parameters are same as those in Sect. 4.5.2. Figure 4.7
shows the uncontrolled displacement responses of the three floors. It is found that
the peak-to-peak oscillation amplitudes of the first, second, and third floors of the
system are 1.4180, 1.5371, and 1.6181 m, respectively. Clearly, to guarantee the
safety of the offshore platform, such a large vibration should be suppressed.
Now, we consider the robust sliding mode control scheme. By Corollary 4.3, the
gain matrix of an RSMC is obtained as
KRSMC = 105 × [−2.9627 1.2771 4.4707 − 0.3213 − 1.5013 − 3.1483].
Depicted in Fig. 4.8 are the responses of the three floors and the required control
force when the RSMC is used to system (2.47), and presented by Fig. 4.9 is
the corresponding curve of the sliding function s(t). It can be obtained through
calculation that under the RSMC, the peak-to-peak oscillation amplitudes of the
first, second, and third floors are reduced from 1.4180, 1.5371, and 1.6181 to 0.2270,
0.2433, and 0.2549 m, respectively, and the control force peak to peak is about
4.5180 × 106 N.
Let γ = 0.15. By Corollary 4.4, we design an RHIC with the gain as
KRHIC = 106 × [−1.0724 0.3596 0.8762 − 0.1348 − 0.3914 − 0.8820].
Under the RHIC, the peak-to-peak oscillation amplitudes of the first, second, and
third floors are 0.1934, 0.2096, and 0.2230 m, respectively; and the range of the
control force is about 8.5366 × 106 N, which can be observed from Fig. 4.10.
Then, in the case of γ = 0.15, by Proposition 4.4, we obtain the gain of an
RSMHC as
KRSMHC = 104 × [−3.8948 0.9946 5.9796 − 0.2226 − 1.2842 − 3.1373].


1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
0 50 100 150 0 50 100 150
Time (s) Time (s)
0.5
-0.5
-1
0 50 100 150
Time (s)
Fig. 4.7 Responses of the three floors of the uncertain system (2.47) without control
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 6
0.2 4
Control force (N)
0.1 2
0 0
-0.1 -2
-0.2 -4
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.8 Responses of the three floors of the uncertain system (2.47) and the control force by
RSMC
8000
6000
4000
2000
-2000
-4000
-6000
-8000
0 50 100 150
Time (s)
Fig. 4.9 Variation of the sliding surface under the RSMC

0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 6
0.2 5
Control force (N)
0.1
0 0
-0.1
-0.2 -5
0 50 100 150 0 50 100 150
Time (s) Time (s)
RHIC

0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 6
0.2
Control force (N)

1
0.1
0 0
-0.1
-1
-0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
RSMHC
When the RSMHC is applied to the uncertain system, the peak-to-peak oscillation
amplitudes of the three floors are reduced to 0.1985, 0.2152 and 0.2278 m, and the
required control force peak to peak is about 1.9309 × 106 N. Figure 4.11 shows
the responses of the three floors of the uncertain system and the curve of the
control force, and Fig. 4.12 depicts the variation of the sliding surface s(t) under
the RSMHC.
For the sake of comparison, the oscillation amplitudes of the uncertain system
and the control force under the RSMC, RHIC, RSMHC, and no control are
summarized in Table 4.3, where for different values of γ , the simulation results
are compared between the RHIC and the RSMHC.
It is found from Table 4.3 that for the uncertain system (2.47) with (4.1),
applying the RSMC, RHIC, and RSMHC, the average oscillation amplitudes of
the three floors are reduced to 16%, 14%, and 14% of the oscillation amplitudes
of the system without control, respectively. In fact, under the RSMC, the average
oscillation amplitude of the three floors is about 0.24 m, while under the RHIC and
the RSMHC, the average oscillation amplitude of the three floors is about 0.22 m,
which indicates that the controlled average oscillation amplitude of the system under
the RSMHC is smaller than the one under the RSMC and almost the same as that
under the RHIC. In addition, it can be observed that the required control force under
the RSMHC is smaller than the ones under the RSMC and the RHIC.
4.6 Conclusions 67
2000
1500
1000
500
-500
-1000
-1500
-2000
0 50 100 150
Time (s)
Fig. 4.12 Variation of the sliding surface under the RSMHC
Table 4.3 Maximum control forces and oscillation amplitudes of three floors of the uncertain
system (2.47) under different controllers
γ Controller Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (106 N)
– No control 1.4180 1.5371 1.6181 –
– RSMC 0.2270 0.2433 0.2549 4.5180
RHIC 0.1934 0.2096 0.2230 8.5366
0.15
RSMHC 0.1985 0.2152 0.2278 1.9309
RHIC 0.1993 0.2159 0.2286 2.1350
0.20
RSMHC 0.2027 0.2199 0.2326 1.7424
RHIC 0.2000 0.2160 0.2292 8.8975
0.25
RSMHC 0.2031 0.2202 0.2324 2.2739
RHIC 0.2014 0.2176 0.2306 2.6955
0.30
RSMHC 0.2041 0.2214 0.2339 1.9714
RHIC 0.2030 0.2193 0.2322 1.9707
0.45
RSMHC 0.2079 0.2251 0.2383 1.6647
4.6 Conclusions
This chapter presents a sliding mode H∞ control scheme for offshore platforms
subject to nonlinear wave forces and external disturbance. Applying this scheme,
controllers for the nominal system and the uncertain system have been designed
to reduce the oscillation amplitudes of the offshore platforms. Based on simulation
results, the designed sliding mode H∞ controller, sliding mode controller, and the
traditional H∞ controller are compared from two aspects: the oscillation amplitudes
of offshore platforms and the control cost. Simulation results have shown that the
proposed sliding mode H∞ control scheme is effective to improve performance of

the offshore platforms. In fact, the controlled oscillation amplitudes of the platforms
under the sliding mode H∞ controller are less than those under the sliding mode
controller and the H∞ controller. In addition, the control force required by the
sliding mode H∞ controller is less than the one by the sliding mode controller and
the H∞ controller.
4.7 Notes
The main results of this chapter are derived from Zhang et al. [90], where both
parametric perturbations of the system and the external disturbance are consid-
ered. Specifically, if the external disturbance is ignored, the offshore platform
system (2.50) reduces to the one as (2.45) by Terro et al. [72] and Zribi et al. [65];
and the uncertain dynamic model (2.47) reduces to the one as (2.49), similar analysis
about the uncertain model can be found in Zhang et al. [99]. In this situation, an
integral sliding mode control scheme and a robust integral sliding mode control
scheme are developed for the nominal system (2.45) and the uncertain system (2.49),
respectively. For more discussions on integral sliding mode control for the offshore
platforms, one refers to [99]. It is found that the integral sliding mode control
schemes can reduce the internal oscillations of the offshore steel jacket platform
dramatically; and the performance of the offshore steel jacket platform under the
integral sliding mode control schemes is better than the ones under the nonlinear
control scheme [65] and the dynamic output feedback control scheme [100]. More
recently, Nourisola et al. [102, 103] have developed adaptive integral sliding mode
control schemes for the offshore steel jacket platforms. For an offshore platform
with an AMD mechanism, Zhang et al. [98] offer an optimal sliding mode controller,
which is better than the feedforward and feedback optimal controller proposed by
Ma et al. [81].
Note that the sliding mode control methods aforesaid are mainly based on the
simplified dynamic models of offshore steel jacket platform [65, 72, 79], where
either only the most dominant mode or the first and the second vibration modes of
the offshore platform with a single load are considered. However, offshore platforms
are of characteristics of combinations of multi-vibration modes. Moreover, offshore
platforms are usually subject to parametric perturbations, nonlinear dynamics, and
mixed effects of external disturbance, such as winds and waves, or earthquake, and
waves and flow. Therefore, it is significant to develop a nonlinear model for an
offshore platform such that its overall performance can be improved. In this sense,
a challenging problem is to establish a proper mathematical model that can exactly
reflect the dynamics of an offshore platform.
In fact, on the one hand, as an efficient, simple and robust control scheme,
sliding mode control is worthy to be further investigated for the complicated
nonlinear dynamic models of jacket-type offshore platform and even other types
of offshore platforms; on the other hand, note that neural networks are suitable for
4.7 Notes 69
approximating nonlinear systems and fuzzy logics are efficient to handle systematic
nonlinearities and uncertainties and easy to implement for structural systems.
Offshore platforms are typical nonlinear systems. However, except for several work
by Zhou and Zhao [92], Chang et al. [93], Kim [94, 95], and Cui and Hong
[96], the study on intelligent structure and/or intelligent controllers has not been
adequately explored for offshore platforms. Specifically, since it is very complicated
and impossible for offshore platforms to establish exact dynamic models. In this
situation, based on a great number of experimental and field data, to explore data-
based dynamic models and to develop effective data-based control schemes for
offshore platforms are of significance both in theory and in real implementations,
which is well worth investigation in the near future.
Chapter 5
Delayed Integral Sliding Mode Control
This chapter is concerned with active control for an offshore steel jacket platform
subject to wave-induced force and parameter perturbations. The offshore steel jacket
platform is shown in Fig. 2.2 [65, 72]. The dynamic model under consideration is
given by (2.44). By intentionally introducing a proper time-delay into the control
channel, a novel sliding mode control scheme is proposed. This scheme utilizes
mixed current and delayed states. It is shown through simulation results that this
scheme is more effective in both improving control performance and reducing
control force of the offshore platform than some existing ones such as delay-free
sliding mode control [99], nonlinear control [65], dynamic output feedback control,
and delayed dynamic output feedback control [100]. Furthermore, it is shown that
the introduced time-delay in this scheme can take values in different ranges, while
the corresponding control performance of the offshore platform is almost at the
similar level.
The rest of this chapter is organized as follows. Section 5.1 presents an scheme to
design a sliding mode controller with mixed current and delayed sates. An numerical
algorithm is given in Sect. 5.2 to solve the controller gain matrices. Section 5.3
provides simulation results to demonstrate the effectiveness of the sliding mode
control scheme with mixed current and delayed states, and to investigate the effects
of artificially introduced time-delays on the control for the offshore platform.
Finally, Sect. 5.4 concludes the chapter, and Sect. 5.5 provides a brief note.
5.1 Design of Delayed Robust Sliding Mode Controllers
In this section, for the system (2.44), we first present a new scheme to design the
sliding mode surface with mixed current and delayed states. Then, we provide an
approach to designing a desired sliding mode controller for the offshore platform.

https://doi.org/10.1007/978-981-13-2986-9_5
72 5 Delayed Integral Sliding Mode Control
5.1.1 Integral Sliding Surface Design
The sliding surface function to be designed is in the following form:

t−τ t
s1 (t) = Gx(t) − GBKτ x(θ )dθ − G(A + BK)x(θ )dθ − Gx0 (5.1)
0 0
where G is a 1 × 6 real matrix to be chosen such that GB is nonsingular, K and Kτ

are 1 × 6 real matrices to be determined, and τ > 0 is a given scalar.
As the state trajectory of the system (2.44) enters the sliding mode surface, we
have s1 (t) = 0 and ṡ1 (t) = 0. Then, from (2.44) and (5.1), an equivalent control
law in the sliding mode can be obtained as
ueq (t) = Kx(t) + Kτ x(t − τ ) − (GB)−1 G[ΔA(t)x(t) + Df (x, t)] (5.2)
where ΔA(t) = M̃ F̃ (t)Ñ. Then, the sliding motion can be written as
ẋ(t) = [A + BK + Ē F̃ (t)Ñ]x(t) + BKτ x(t − τ ) + D̄f (x, t) (5.3)
where Ē = ḠM̃ and D̄ = ḠD with Ḡ = I − B(GB)−1 G.

The initial condition of the state x(t) on [−τ, 0] is supplemented as
x(θ ) = ϕ(θ ), θ ∈ [−τ, 0]
where ϕ ∈ W and W denotes the Banach space of absolutely continuous functions

[−τ, 0] → Rn with square-integrable derivative and with the norm
0 0
ϕ2W = ϕ(0)2 + ϕ(ξ )2 dξ + ϕ̇(ξ )2 dξ.
−τ −τ
The following theorem provides a sufficient condition on the existence of

controller gain matrices K̄ and K̄τ such that the sliding motion of (5.3) is robustly
stable.
Theorem 5.1 For given scalars μ > 0 and τ > 0, the sliding motion (5.3) is
robustly stable if there exist 6 × 6 matrices P̄ > 0, Q̄ > 0, R̄ > 0, 1 × 6 matrices
K̄ and K̄τ , and a scalar ε > 0 such that
⎡ ⎤
Ψ11 Ψ12 D̄ Ψ14 μP̄ εĒ P̄ Ñ T
⎢ ∗ 0 ⎥
⎢ Ψ22 0 Ψ24 0 0 ⎥
⎢ ∗ ∗ −I 0 ⎥
⎢ Ψ34 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ Ψ44 0 Ψ46 0 ⎥ < 0 (5.4)
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
5.1 Design of Delayed Robust Sliding Mode Controllers 73
where
Ψ11 = AP̄ + P̄ AT + B K̄ + K̄ T B T + Q̄ − R̄
Ψ12 = B K̄τ + R̄, Ψ14 = τ P̄ AT + τ K̄ T B T
Ψ22 = −Q̄ − R̄, Ψ24 = τ K̄τT B T , Ψ34 = τ D̄ T
Ψ44 = −P̄ R̄ −1 P̄ , Ψ46 = ετ Ē
Moreover, the gain matrices K and Kτ in (5.1) are given by K = K̄ P̄ −1 and

Kτ = K̄τ P̄ −1 , respectively.
Proof Choose a Lyapunov-Krasovskii functional candidate as
t
V1 (xt ) = x T (t)P x(t) + x T (s)Qx(s)ds
t−τ
0 t
+τ ds ẋ T (θ )R ẋ(θ )dθ
−τ t+s
where xt = x(t + s), s ∈ [−τ, 0], P > 0, Q > 0 and R > 0. Taking the
time derivative of V1 (xt ) along the trajectory of (5.3), after some simple algebraic
manipulation, we have
V̇1 (xt ) ≤ x T (t)[P A + AT P + P BK + K T B T P + Q

+ P Ē F̃ (t)Ñ + Ñ T F̃ T (t)Ē T P + μ2 I ]x(t)
+ 2x T (t)P BKτ x(t − τ ) − f T (x, t)f (x, t)
− x T (t − τ )Qx(t − τ ) + 2x T (t)P D̄f (x, t)
t
+ ẋ T (t)(τ 2 R)ẋ(t) − ẋ T (s)(τ R)ẋ(s)ds (5.5)
t−τ
Applying Lemma 2.6 to the integral term in (5.5), together with (5.3), one has

V̇1 (xt ) ≤ ηT (t) Ξ (t) + Γ T (t)(τ 2 R)Γ (t) η(t)
where
⎡ ⎤
U (t) P BKτ + R P D̄
Ξ (t) := ⎣ ∗ −Q − R 0 ⎦
∗ ∗ −I
Γ (t) := [A + BK + Ē F̃ (t)Ñ BKτ D̄]
T
η(t) := x T (t) x T (t − τ ) f T (x, t)
with
U (t) = P A + AT P + P BK + K T B T P + Q − R
+μ2 I + P Ē F̃ (t)Ñ + Ñ T F̃ T (t)Ē T P
If Ξ (t) + Γ T (t)(τ 2 R)Γ (t) < 0, then there exists a scalar κ > 0 such that V̇1 (xt ) ≤
−κx T (t)x(t) < 0 for x(t) = 0, which ensures the asymptotic stability of the sliding
mode in (5.3). Fortunately, Ξ (t) + Γ T (t)(τ 2 R)Γ (t) < 0 is implied by (5.4). To
show that, by Lemma 2.1
Ξ (t) + Γ T (t)(τ 2 R)Γ (t) < 0

Ξ (t) τ Γ T (t)R
⇐⇒ Ω1 := <0
∗ −R
Notice that Ω can be rewritten as
Ω1 = Q + H F̃ (t)E + ET F̃ (t)T H T
where
⎡ ⎤
Ũ P BKτ + R P D̄ τ (A+BK)T R
⎢ ∗ −Q − R 0 τ (BKτ )T R ⎥
Q := ⎢
⎣∗
⎥
⎦
∗ −I τ D̄ T R
∗ ∗ ∗ −R
H := col{P Ē, 0, 0, τ R Ē}
E := [Ñ 0 0 0 ]
Ũ := P A+AT P +P BK +K T B T P +Q−R+μ2 I
By Lemma 2.2, Ω1 < 0 if and only if there exists a scalar ε > 0 such that
Q + ε−1 ET E + εH H T < 0
which, by Schur complement, is equivalent to

⎡ ⎤
Υ11 Υ12 P D̄ Υ14 μI P Ē Ñ T
⎢ ∗ 0 ⎥
⎢ Υ22 0 Υ24 0 0 ⎥
⎢ ∗ ∗ −I 0 ⎥
⎢ Υ34 0 0 ⎥
⎢ ⎥
Ω2 := ⎢ ∗ ∗ ∗ −R 0 Υ46 0 ⎥<0
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ Υ66 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
5.1 Design of Delayed Robust Sliding Mode Controllers 75
where
Υ11 = P A + AT P + P BK + K T B T P + Q − R,
Υ12 = P BKτ + R, Υ14 = τ AT R + τ K T B T R,
Υ22 = −Q − R, Υ24 = τ KτT B T R,
Υ34 = τ D̄ T R, Υ46 = τ R Ē, Υ66 = −ε−1 I.
Pre- and post-multiply Ω2 by diag{P −1 , P −1 , I, R −1 , I, εI , I } and its transpose,

respectively, and let P̄ = P −1 , Q̄ = P −1 QP −1 , R̄ = P −1 RP −1 , K̄ = KP −1 ,
and K̄τ = Kτ P −1 . Then we arrive at the condition (5.4). Thus, if there exist P >
0, Q > 0,R > 0, K and Kτ such that (5.4) is satisfied, then the inequality Ξ (t) +
Γ T (t)(τ 2 R)Γ (t) < 0 is true. The proof is thus completed.

5.1.2 Design of a Robust Delayed Sliding Mode Controller
Now, we proceed to design a robust sliding mode controller with mixed current and
delayed states such that the reachability of the specified sliding surface s1 (t) = 0 is
ensured.
Based on the obtained controller gain matrices K and Kτ in Theorem 5.1,
a robust sliding mode controller with mixed current and delayed states can be
designed as
u(t) = Kx(t) + Kτ x(t − τ ) − ρ1 (x(t))sgn(s1 (t)) (5.6)
where sgn(·) is the sign function, and the switching gain ρ1 (x(t)) is given by
ρ1 (x(t)) = (GB)−1 [η + (η0 + μ GD) x(t)] (5.7)

! ! ! !
! ! ! !
with η > 0 and η0 = !GM̃ ! · !Ñ !.
The theorem below will show that the controller (5.6) can drive the state
trajectory of system (2.44) into the sliding surface s1 (t) = 0 in finite time.
Theorem 5.2 Under the robust sliding mode controller (5.6), the state trajectory
of system (2.44) can be driven into the sliding surface s1 (t) = 0 in finite time and
maintain on it thereafter.
Proof Choose a Lyapunov function as
1 2
V2 (s1 (t)) = s (t) (5.8)
2 1
From (2.44), (5.1) and (5.6), together with (5.7), we have
s˙1 (t) = GM̃ F̃ (t)Ñx(t) + GDf (x, t) − [η + (η0 + μ GD) x(t)] sgn(s1 (t))
(5.9)
Therefore, from (2.43), (5.8) and (5.9), we obtain
V̇2 (s1 )(t) ≤ −η|s1 (t)|, ∀s1 (t) = 0 (5.10)
which implies that under the controller (5.6), the sliding surface s1 (t) = 0 is
reachable in finite time and the state trajectory of system (2.44) maintains on it
thereafter. This completes the proof.

5.1.3 Design of a Delayed Sliding Mode Controller
If there is no uncertainty in the system’s matrices, then the system (2.44) reduces
to (2.45). In this case, we can design a sliding mode controller with mixed current
and delayed states as
u(t) = Kx(t) + Kτ x(t − τ ) − ρ2 (x(t))sgn(s1 (t)) (5.11)
where s1 (t) is defined by (5.1), and
ρ2 (x(t)) = (GB)−1 [η + μ GD x(t)] (5.12)
By Theorems 5.1 and 5.2, we have the following result.

Corollary 5.1 For given scalars μ > 0 and τ > 0 and controller (5.11),
system (2.45) is asymptotically stable in the sliding surface s1 (t) = 0, the sliding
surface is reachable in finite time, and the state trajectory of system (2.45) maintains
on it thereafter, if there exist 6 × 6 matrices P̄ > 0, Q̄ > 0, R̄ > 0, 1 × 6 matrices
K̄ and K̄τ such that
⎡ ⎤
Ψ11 Ψ12 D̄ Ψ14 μP̄
⎢ ∗ 0 ⎥
⎢ Ψ22 0 Ψ24 ⎥
⎢ ⎥
⎢ ∗ ∗ −I Ψ34 0 ⎥<0 (5.13)
⎢ ⎥
⎣ ∗ ∗ ∗ Ψ44 0 ⎦
∗ ∗ ∗ ∗ −I
Moreover, the gain matrices K and Kτ in (5.11) are given by K = K̄ P̄ −1 and

Kτ = K̄τ P̄ −1 , respectively.
Remark 5.1 If Kτ = 0, then (5.6) and (5.11) immediately reduce to the ones in [99],
respectively. It means that the robust sliding mode controller and the sliding mode
controller considered in [99] are the special case of the controllers (5.6) and (5.11),
respectively.
Remark 5.2 Notice that the sliding mode control may possibly exhibit some
chattering behaviors. When the sliding mode controller is applied to the offshore
5.2 A Computational Algorithm 77
platform, one way to avoid chattering is to replace sgn(s1 (t)) with the follow-
ing sigmoid-like function to “smooth” the behaviors of the controller in (5.6)
and (5.11)
s1 (t)
χ (s1 (t)) = (5.14)
|s1 (t)| + δ
where δ is a small positive scalar [135, 136].
5.2 A Computational Algorithm
Notice that from (5.4) and (5.13) the matrix inequalities are nonlinear due to the
block element Ψ44 . To solve the gain matrices K and Kτ from (5.4) or (5.13),
we now present a numerical algorithm by employing the cone complementary
linearization approach [137]. Since inequalities (5.4) and (5.13) are of the same
structure, we only focus on the calculation of the gain matrices K and Kτ
from (5.13).
Introduce a new matrix S > 0 such that S ≤ P̄ R̄ −1 P̄ , which is equivalent to

S −1 P̄ −1
≥0 (5.15)
∗ R̄ −1
Let S̄ = S −1 , L̄ = P̄ −1 and M̄ = R̄ −1 . Then the nonlinear matrix inequality (5.13)

holds if
⎡ ⎤
Ψ11 Ψ12 D̄ Ψ14 μP̄
⎢ ∗ Ψ ⎥
⎢ 22 0 Ψ24 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I Ψ34 0 ⎥ < 0 (5.16)
⎢ ⎥
⎣ ∗ ∗ ∗ −S 0 ⎦
∗ ∗ ∗ ∗ −I

S̄ L̄
≥0 (5.17)
∗ M̄
S̄S = I, L̄P̄ = I, M̄ R̄ = I (5.18)
The problem formulated by the conditions (5.16), (5.17), and (5.18) is a non-
convex feasibility problem (NCFP) due to the equality constraints in (5.18). In
the sequel, we convert the NCFP formulated by (5.16), (5.17), and (5.18) to
the following nonlinear minimization problem subject to a set of linear matrix
inequalities.
A nonlinear minimization problem (NMP)
Minimize Tr (S̄S + L̄P̄ + M̄ R̄) (5.19)

Subject to (5.16), (5.17) and

S̄ I L̄ I M̄ I
≥ 0, ≥ 0, ≥0 (5.20)
∗S ∗ P̄ ∗ R̄
To solve the above NMP, we give an algorithm. For this goal, we introduce some
notations as
" #
Γ = P̄, Q̄, R̄, S̄, S, L̄, M̄, K̄τ , K̄
(0)
Γ (0) = P̄ (0) , Q̄(0) , R̄ (0) , S̄ (0) , S (0) , L̄(0) , M̄ (0) , K̄τ , K̄ (0)
Algorithm 1 An iterative algorithm to solve the NMP

Step 1 Find a feasible solution satisfying (5.16), (5.17) and (5.20), set k = 0, and
specify a small ! > 0;
Step 2 Solve the NMP below on the matrix variable set Γ
" #
Minimize Tr S̄ (k) S + S (k) S̄ + L̄(k) P̄ + P̄ (k) L̄ + M̄ (k) R̄ + R̄ (k) M̄
Subject to (5.16), (5.17), and (5.20).
Set S̄ (k+1) = S̄, S (k+1) = S, L̄(k+1) = L̄, P̄ (k+1) = P̄ , M̄ (k+1) =

M̄, R̄ (k+1) = R̄;
Step 3 If one of the matrix inequalities (5.13) and
$ " # $
$Tr S̄ (k) S + S (k) S̄ + L̄(k) P̄ + P̄ (k) L̄ + M̄ (k) R̄ + R̄ (k) M̄ − 36$ < !
is not satisfied within a specified number of iterations, then exit; otherwise, set
k = k + 1 and go to Step 2.
In this section, we will show through simulation the effectiveness of the sliding
mode control scheme with mixed current and delayed states for system (2.45) and
the robust sliding mode control scheme with mixed current and delayed states for
uncertain system (2.44), respectively.
An offshore platform with a TMD mechanism presented in Fig. 2.2 is simulated.
The parameters of the system and the waves are given in Table 4.1. With the setting
in the table, the nonlinear wave force f (x, t) can be computed as Appendix A of
[65], and the matrices A and B in (2.44) and (2.45) can be obtained as (4.34).
5.3.1 Simulation Results for the Nominal System
In this subsection, based on Algorithm 1, we first design a sliding mode controller

with mixed current and delayed states (SMC-MCDS) in the form (5.11) for the
nominal system (2.45) and compare the SMC-MCDS with the delay-free sliding
mode controller (SMC) in [99] from the controlled oscillation amplitudes of
the offshore platform and the size of control force. Then, we investigate the
effects of time-delays on the sliding mode control for the offshore platform.
Third, we compare the SMC-MCDS with a nonlinear controller (NLC) [65],
a dynamic output feedback controller (DOFC), and a delayed dynamic output
feedback controller (DDOFC) [100], respectively, to show some advantages of the
SMC-MCDS.
5.3.1.1 Comparison of SMC-MCDS and SMC
We consider the sliding mode control scheme with mixed current state and delayed
state. Set G = [1000 10 1000 1 1000 1], μ = 0.8, η = 0.1, δ = 0.005, and
τ = 0.01 s. By Algorithm 1, an SMC-MCDS of the form (5.11) can be obtained,
and the gain matrices K and Kτ are given as
K = [−1184.7 175.1 1709.5 51.3 −231 −894.8]

Kτ = [−672.44 −179.86 −270.62 580.32 404.52 −727.44]
Under SMC-MCDS, the response curves of the three floors and the
control force are presented in Fig. 5.1. It can be computed that the oscillation
amplitudes peak to peak of the first, second, and third floors are 0.1868, 0.2041,
and 0.2182 m, respectively, and the required peak to peak control force is
1.3745 × 105 N.
Then, under SMC, it can be seen from [99] that the oscillation amplitudes of
the first, second, and third floors are 0.2192, 0.2301 and 0.2383 m peak to peak,
respectively, and the control force required is about 2.1565 ×105 N. The oscillation
amplitudes of the three floors and the control force under SMC-MCDS and SMC
are listed in Table 5.1, which indicates that compared with SMC, SMC-MCDS has
some significant advantages as follows.
• The oscillation amplitudes of the three floors under SMC-MCDS are less than
those under SMC.
• The control force required by SMC-MCDS is smaller than the one by SMC. In
fact, the control force required by SMC is about 1.57 times as that by SMC-
MCDS.
0.2 0.2

0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
5
x 10
0.2 1
Control force (N)

0.1 0.5
0 0
−0.1 −0.5
−0.2 −1
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
Fig. 5.1 Responses of the three floors of the system and the required control force under SMC-
MCDS
Table 5.1 The maximum oscillation amplitudes of three floors and control force under SMC and
SMC-MCDS
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (105 N)
SMC [99] 0.2192 0.2301 0.2383 2.1565
SMC-MCDS 0.1868 0.2041 0.2182 1.3745
5.3.1.2 The Effects of Time-Delays on Sliding Mode Control
Under SMC-MCDS, for different values of time-delay τ , the maximum oscillation

amplitudes of the three floors and the control force are given by Table 5.2. It is
clear that as time-delay changes from 0.01 to 0.15 s, or from 0.43 to 0.56 s, both
the maximum oscillation amplitudes of the three floors and the required control
force by SMC-MCDS are much less than those by SMC, which can be observed
from Figs. 5.2 and 5.3. It shows that in the ranges of 0.01–0.15 and 0.43–0.56 s,
time-delays have positive effects on the sliding mode control for the offshore
platform.
Table 5.2 The oscillation amplitudes of three floors and control force under SMC-MCDS for
different time-delays
Time-delay (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (105 N)
0.01 0.1868 0.2041 0.2182 1.3745
0.07 0.1857 0.2035 0.2183 1.3605
0.13 0.1846 0.2029 0.2198 1.3655
0.15 0.1895 0.2016 0.2288 2.0296
0.16 0.2592 0.2241 0.2833 7.3935
0.42 0.3135 0.2579 0.3047 5.4760
0.43 0.1883 0.2037 0.2255 1.4739
0.48 0.1819 0.2024 0.2224 1.3781
0.54 0.1918 0.2103 0.2260 1.5044
0.56 0.2023 0.2201 0.2335 1.7694
0.57 0.2141 0.2322 0.2445 2.0809
0.60 0.3084 0.3246 0.3265 4.8688
0.25 0.26
0.24 SMC SMC
Amplitudes of floor 1 (m)
SMC−MCDS 0.25 SMC−MCDS

0.23
0.24
0.22
0.21 0.23
0.2
0.22
0.19
0.21
0.18
0.17 0.2
0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.15 0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.15
Time−delay (s) Time−delay (s)
5
x 10
0.26 2.7
SMC SMC
Maximum control force (N)
2.5
0.25 SMC−MCDS SMC−MCDS
2.2
0.24
1.9
0.23
0.22 1.6
0.21 1.3
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.15
Fig. 5.2 Oscillation amplitudes of the system and the control force under SMC and SMC-MCDS
as a time-delay is in the range of 0.01–0.15 s
0.25 0.25
0.24 SMC SMC

SMC−MCDS 0.24 SMC−MCDS
0.23
0.23
0.22
0.21 0.22
0.2
0.21
0.19
0.2
0.18
0.17 0.19
0.43 0.45 0.47 0.49 0.51 0.53 0.56 0.43 0.45 0.47 0.49 0.51 0.53 0.56
5
x 10
0.26 3
SMC SMC

SMC−MCDS SMC−MCDS
0.25 2.5
0.24 2
0.23 1.5
0.22 1
0.43 0.45 0.47 0.49 0.51 0.53 0.56 0.43 0.45 0.47 0.49 0.51 0.53 0.56
Fig. 5.3 Oscillation amplitudes of the system and the control force under SMC and SMC-MCDS
as a time-delay is in the range of 0.43–0.56 s
Table 5.3 The oscillation amplitudes of three floors and control force under NLC, DOFC,
DDOFC, and SMC-MCDS
NLC [65] 0.3050 0.3050 0.3050 2.0
DOFC [100] 0.2329 0.2543 0.2705 4.0
DDOFC [100] 0.2034 0.2232 0.2391 0.6
SMC-MCDS 0.1868 0.2041 0.2182 1.4
5.3.1.3 Comparison of SMC-MCDS and Some Other Existing Controllers
Now, we compare SMC-MCDS with NLC [65], DOFC, and DDOFC [100],
respectively. Table 5.3 presents the maximum oscillation amplitudes of the three
floors and the required control force under the controllers mentioned above.
On the one hand, it can be seen from Table 5.3 that under NLC and DOFC,
the average oscillation amplitudes of the three floors are 0.3050 and 0.2526 m,
respectively, while under the SMC-MCDS, the average vibration amplitude is
reduced to about 0.2030 m. Moreover, the required control force by NLC and DOFC
is 1.4 times and 2.9 times as that by SMC-MCDS, respectively. It shows that
compared with NLC and DOFC, SMC-MCDS is capable of taking better vibration
reduction of the offshore platform at a much less control force.
On the other hand, it can also be seen from Table 5.3 that DDOFC requires less
control force leading to almost the same oscillation amplitudes of the three floors
as those under the SMC-MCDS. However, from the implementation point of view,
the DDOFC is difficult to be implemented because the introduced time-delay should
be less than 0.11 s [100], while the physical time-delay is usually larger than 0.14 s
[138]. Nevertheless, the SMC-MCDS provides a larger range from 0.43 to 0.56 s
for the introduced time-delay to be chosen, which can be seen from Table 5.2 and
Figs. 5.2 and 5.3.
5.3.2 Simulation Results for the Uncertain System
This subsection will focus on designing a robust sliding mode controller with mixed
current and delayed states (RSMC-MCDS) for the uncertain system (2.44). Some
comparisons between the designed RSMC-MCDS and the delay-free robust sliding
mode controller (RSMC) [99] and the robust dynamic output feedback controller
(RDOFC) will be made from the controlled oscillation amplitudes of the offshore
platform and the size of control force required. Then, the effects of time-delays on
the robust sliding mode control for the offshore platform will be investigated.
Suppose that ξ̂i = 0.02 (i = 1, 2, T ), ĈT = 0.05, K̂T = 1. Then from (2.26),
one has ω̂T = 0.0003. From (2.34), it is clear that

0 0 0 0
Ē0 = , Ēi = , i = 1, 2
1 0.05 0.001 0.02

0 0 −6.6102 −0.5454
ĒT = , HT =
0.0003 0.02 0 −3.6360

−6.6102−0.0182 −236.24 −0.1087
H̄1 = , H̄2 =
0 −3.6360 0 −21.737
The time-varying perturbation function is supposed to be F̃ (t) = sin(t) · I .
5.3.2.1 Comparison of RSMC-MCDS and Some Other Existing

Controllers
To begin with, when no controller is acted on the uncertain system (2.44), Fig. 5.4
depicts the response curves of the three floors of the system. The oscillation
amplitudes peak to peak of the first, second, and third floors of the system are
1.4284, 1.5479, and 1.6252 m, respectively.
Then, under RDOFC [100], it can be computed that the oscillation amplitudes
peak to peak of the three floors are 0.7315, 0.7991, and 0.8504 m, respectively, and
the required control force peak to peak is about 1.4166 ×106 N.

0.5 0.5
0 0
−0.5 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
1
0.5
−0.5
−1
0 20 40 60 80 100
Time (s)
Fig. 5.4 Responses of the three floors of the uncertain system without control
Third, we consider the delay-free robust sliding mode control scheme. Set μ =
0.8, η = 0.1, and G = [1000 10 1000 1 1000 100]. By Proposition 1 in [99], an
RSMC can be obtained, whose gain K is given by
K = [−49610 14960 15264 − 2920 − 25600 − 51610]
Under RSMC, the response curves of the three floors and the control force are
plotted in Fig. 5.5. From this figure, the oscillation amplitudes peak to peak of the
first, second, and third floors are 0.2251, 0.2445, and 0.2581 m, respectively, and
the control force required is about 1.5942 × 106 N.
Finally, we turn to the robust sliding mode control scheme with mixed current and
delayed states. Let τ = 0.01 s. Solve the matrix inequality (5.4) to get an RSMC-
MCDS of the form (5.6) with

K = −2258.4 344.3 6403.4 309.4 −422.8 −1965.2

Kτ = −386.6 2.51 181.5 321.6 154.9 −349.8
Under RSMC-MCDS, the response curves of the three floors and the control force
are depicted in Fig. 5.6. The oscillation amplitudes of the three floors and the control
force required are listed in Table 5.4. From the table, one can see that the oscillation
amplitudes of the three floors are greatly reduced from 1.4284, 1.5479, and 1.6252
0.2 0.2

0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
6
x 10
0.2 1
0.1 0.5
Control force (N)

0 0
−0.1 −0.5
−0.2 −1
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
Fig. 5.5 Responses of the three floors of the uncertain system under RSMC
0.2 0.2
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
6
x 10
0.2 1
0.1 0.5
Control force (N)
0 0
−0.1 −0.5
−0.2 −1
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
Fig. 5.6 Responses of the three floors of the uncertain system under RSMC-MCDS
Table 5.4 The oscillation amplitudes of three floors and control force under no control, RDOFC,
RSMC, and RSMC-MCDS
No Control 1.4284 1.5479 1.6252 –
RDOFC [100] 0.7315 0.7991 0.8504 1.4166
RSMC [99] 0.2251 0.2445 0.2581 1.5942
RSMC-MCDS 0.2237 0.2427 0.2572 1.0459
Table 5.5 The oscillation amplitudes of three floors and control force under RSMC-MCDS for
different time-delays τ
τ (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (106 N)
0.01 0.2237 0.2427 0.2572 1.0459
0.03 0.2239 0.2436 0.2577 1.0410
0.05 0.2242 0.2438 0.2579 1.0415
0.07 0.2240 0.2438 0.2580 1.0403
0.08 0.2244 0.2441 0.2584 1.0623
0.10 0.2249 0.2443 0.2586 1.0623
0.20 0.2268 0.2453 0.2593 1.0711
0.40 0.2283 0.2315 0.2447 1.0552
0.60 0.2291 0.2483 0.2626 1.0653
0.80 0.2293 0.2483 0.2623 1.0644
to 0.2237, 0.2427, and 0.2572 m peak to peak, respectively. In comparison with

control schemes proposed in [100] and [99], one can see that
• The oscillation amplitudes of the three floors under RSMC-MCDS proposed in
this paper are less than those under RDOFC in [100] and RSMC in [99];
• RSMC-MCDS requires less control force than RDFOC and RSMC. In fact, the
control forces by RDOFC and RSMC are 1.354 and 1.524 times as the one by
RSMC-MCDS, respectively.
More specifically, it should be mentioned that one cannot design a robust delayed
dynamic output feedback controller based on Proposition 4 in [100] when the
parameter uncertainties are taken into account in the offshore platform. Thus, the
delayed dynamic feedback control scheme proposed in [100] fails to draw any
conclusion on the safety of the offshore platform with parameter uncertainties.
5.3.2.2 The Effects of Time-Delays on Robust Sliding Mode Control
To investigate the effects of time-delays on robust sliding mode control for the
offshore platform, for various values of the time-delay τ , we calculate the maximum
oscillation amplitudes of the three floors and the control force required under
RSMC-MCDS. The obtained results are listed in Table 5.5. From this table, one
can see that if the time-delay takes values in the range of 0.01–0.80 s, the oscillation
0.231 0.249
RSMC RSMC

0.23 0.248
RSMC−MCDS RSMC−MCDS
0.229
0.247
0.228
0.246
0.227
0.245
0.226
0.225 0.244
0.224 0.243
0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8 0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8
6
x 10
0.264 2

0.263 RSMC RSMC
RSMC−MCDS 1.8 RSMC−MCDS

0.262
0.261 1.6
0.26
0.259 1.4
0.258
1.2
0.257
0.256 1
0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8 0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8
Fig. 5.7 Oscillation amplitudes of the system and the control force under RSMC and RSMC-
MCDS as a time-delay is in the range of 0.01–0.80 s
amplitudes of the three floors under RSMC-MCDS are almost at the same level
as those under RSMC, while the control forces required under RSMC-MCDS are
less than those under RSMC. Specially, if the time-delay τ changes from 0.01 to
0.07 s, both the oscillation amplitudes of the three floors and the control force under
RSMC-MCDS are less than those under RSMC, which is confirmed by Figs. 5.7
and 5.8. In summary, the simulation results show that, by introducing some proper
time-delays into the control channel, the offshore platform can work in a more safe
environment. The proposed sliding mode controller with mixed current and delayed
states has following remarkable characteristics.
• Compared with SMC (RSMC) and DOFC (RDOFC) [100], SMC-MCDS
(RSMC-MCDS) requires much less control force, and the oscillation amplitudes
of the three floors of the offshore platform under SMC-MCDS (RSMC-MCDS)
are less than those under SMC (RSMC) or DOFC (RDOFC).
• Compared with NLC [65] and DOFC [100], it is found that the oscillation
amplitudes of the three floors of the offshore platform under SMC-MCDS are
less than those under NLC or DOFC. Moreover, the control force required by
SMC-MCDS is smaller than that by NLC or DOFC.
• Compared with DDOFC [100], SMC-MCDS provides more options for a time-
delay to be chosen such that the offshore platform works in a more safe
environment.
0.227 0.248
RSMC RSMC

0.247
0.226 RSMC−MCDS RSMC−MCDS
0.246
0.225 0.245
0.224 0.244
0.243
0.223
0.242
0.222 0.241
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.01 0.02 0.03 0.04 0.05 0.06 0.07
6
x 10
0.261 2

RSMC RSMC
0.26 RSMC−MCDS 1.8 RSMC−MCDS

0.259
1.6
0.258
1.4
0.257
0.256 1.2
0.255 1
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Fig. 5.8 Oscillation amplitudes of the system and the control force under RSMC and RSMC-
MCDS as a time-delay is in the range of 0.01–0.07 s
Remark 5.3 If one wants to implement the method proposed in this chapter, the
generalized coordinates z1 and z2 and their derivatives can be derived from the
corresponding displacements and velocities in physical units. The relation between
the generalized coordinates and the physical units can be found in Appendix B in
[72].
Remark 5.4 It should be pointed out that in this chapter, all the current and delayed
states are available for feedback. If some states and delayed states are not available
for feedback, one can consider a state observer to design a suitable delayed sliding
mode controller to attenuate the wave-induced vibration of the offshore platform.
5.4 Conclusions
In this chapter, a general uncertain model of the offshore platform has been
established by taking the parameter perturbations of the offshore platform with the
active tuned mass damper. Based on this uncertain model, a novel integral sliding
mode control scheme with mixed current and delayed states has been proposed for
the offshore platform subject to nonlinear self-excited wave force and parameter
perturbations. Then the sliding mode controller design has been cast to a convex
5.5 Notes 89
optimization problem with linear matrix inequality constraints. The effects of time-
delays on sliding mode control for an offshore steel jacket platform have been
investigated. As a result, by properly introducing the time-delays into the control
channel, the obtained sliding mode controller can significantly reduce both internal
oscillations of the offshore platform and required control force, which have been
confirmed by simulation results.
5.5 Notes
The delayed feedback control scheme is first applied to the active control for
offshore platform system by Zhang, Han and Han [100], where a delayed dynamic
output feedback controller is designed, and the positive effects of introduced
time-delays on dynamic output feedback control for the system are investigated.
However, the designed time-delay is too small from the point of view of controller
implementation. To solve this problem, the delayed sliding mode controller with
mixed current and delayed states are provided in this chapter, which is mainly
taken from [101]. An in-depth study of such an idea using delayed sliding mode
control can be found in Nourisola and Ahmadi [102] and Nourisola, Ahmadi, and
Tavakoli [103], where the delayed adaptive sliding mode control scheme are studied.
In [107], by intensionally introducing a time-varying delay into the control channel,
a delayed robust sliding mode H∞ controller is designed to reduce vibration of the
offshore platform subject to self-excited nonlinear wave force, external disturbance
and parametric uncertainties. Simulation results show that the delayed robust sliding
mode H∞ controller is better than the traditional robust sliding mode controller and
robust sliding mode H∞ controller [90].
From the reducing vibration of offshore platforms and saving the control cost
point of view, delayed sliding mode control feedback control strategies are effective
and have some distinct advantages. However, it is a key point to choose proper
time-delays and thereby to guarantee the control performance of offshore platforms.
Consequently, a critical value, an optimal value and/or interval of time-delays with
positive effects on control performance of offshore platforms should be further
investigated. Another issue of delayed sliding mode control is that if some states and
delayed states are not available, necessary alternatives deserve deeper investigation.
Chapter 6
Delayed Robust Non-fragile H∞ Control
This chapter is concerned with a delayed non-fragile H∞ control scheme for an

offshore steel jacket platform subject to self-excited nonlinear hydrodynamic force
and external disturbance. By intentionally introducing a time-delay into the control
channel, a delayed robust non-fragile H∞ controller is designed to reduce the
vibration amplitudes of the offshore platform. The positive effects of time-delays
on the non-fragile H∞ control for the offshore platform are investigated. It is
shown through simulation results that (i) the proposed delayed non-fragile H∞
controller is effective to attenuate vibration of the offshore platform, (ii) the control
force required by the delayed non-fragile H∞ controller is smaller than the one by
the delay-free non-fragile H∞ controller, and (iii) the time-delays can be used to
improve the control performance of the offshore platform.
The rest of this chapter is organized as follows. A delayed non-fragile H∞ control
problem for an offshore platform subject to parameter uncertainties, self-excited
nonlinear hydrodynamic force, and external disturbance is presented in Sect. 6.1.
Section 6.2 proposes some sufficient conditions of the existence of the delayed non-
fragile H∞ controller with mixed current and delayed states. In Sect. 6.3, simulation
results are given to compare the delayed non-fragile H∞ control scheme with the
delay-free non-fragile H∞ control scheme from two aspects: controlled oscillation
amplitudes and required control cost. As a special case of the delayed non-fragile
H∞ control scheme, the delayed state feedback control scheme is compared with the
nonlinear control scheme [65], the delayed output feedback control scheme [100],
and an integral sliding mode control scheme [99], respectively. Finally, Sect. 6.4
concludes the chapter, and Sect. 6.5 presents a brief note.

https://doi.org/10.1007/978-981-13-2986-9_6
92 6 Delayed Robust Non-fragile H∞ Control
In this chapter, the offshore platform under consideration is shown in Fig. 2.2
[65, 72]. The dynamic model of the offshore platform subject to the self-excited
hydrodynamic forces and the external disturbance is given (2.47). It is assumed that
the external disturbance ζ (t) ∈ L2 [0, ∞).
The control output equation is given by (4.1). To design a delayed robust non-
fragile H∞ control law with mixed current and delayed states such that under the
designed control law, the system (2.47) with ζ (t) = 0 is robustly stable; and under
the zero initial condition, the H∞ performance
η(t) ≤ γ ζ (t) (6.1)
of the resulting closed-loop system is guaranteed for nonzero ζ (t) and a prescribed
γ > 0.
The delayed robust non-fragile control law is designed as
u(t) = [K1 + ΔK1 (t)]x(t) + [K2 + ΔK2 (t)]x(t − h) (6.2)
where K1 and K2 are gain matrices to be determined, h > 0 is the intentionally

introduced time-delay, and

ΔK1 (t) ΔK2 (t) = V G(t) U1 U2 (6.3)
where V , U1 , and U2 are known matrices with appropriate dimensions and G(t) is
an unknown time-varying matrix satisfying
GT (t)G(t) ≤ I, ∀t > 0 (6.4)
6.2 Design of a Delayed Robust Non-fragile H∞ Controller
In this section, some sufficient conditions for the existence of the delayed robust
non-fragile H∞ controllers are developed.
Substituting Eq. (6.2) into Eq. (2.47) yields
ẋ(t) = A(t)x(t) + B(t)x(t − h) + Df (x, t) + D0 ζ (t), x(0) = x0 (6.5)
where
A(t) = A + BK1 + M̂ F̂ (t)N̂ + BV G(t)U1

B(t) = BK2 + BV G(t)U2
6.2 Design of a Delayed Robust Non-fragile H∞ Controller 93
The following proposition provides a sufficient condition for the existence of the
delayed robust non-fragile control law (6.2).
Proposition 6.1 Under Assumption 1 of Chap. 4, for given scalars γ > 0, μ > 0,
and h > 0, the closed-loop system (6.5) with ζ (t) = 0 is robustly stable, and the
H∞ performance (6.1) is guaranteed, if there exist 6 × 6 matrices P > 0, Q > 0,
R > 0, M1 , M2 , Z1 ≥ 0, Z2 , Z3 ≥ 0, 1 × 6 matrices K1 and K2 , and scalars ε > 0
and ! > 0 such that
⎡ ⎤
Ψ11 Ψ12 P D P D0 Ψ15T C1T μI εP BV U1T !P M̂ N̂ T
⎢ ∗ T U2T 0 ⎥
⎢ Ψ22 0 0 Ψ25 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I 0 DT 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ −γ 2 I D0T D1T 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ Ψ55 0 0 εBV 0 ! M̂ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥<0 (6.6)
⎢ ∗ ∗ ∗ ∗ ∗ ∗ −I 0 ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
and
⎡ ⎤
R M1 M2
⎣ ∗ Z1 Z2 ⎦ ≥ 0 (6.7)
∗ ∗ Z3
where
⎧
⎨ Ψ11 = P A + AT P + P BK1 + K1T B T P + Q + M1 + M1T + hZ1
Ψ = P BK2 − M1T + M2 + hZ2 , Ψ15 = A + BK1 (6.8)
⎩ 12
Ψ22 = −Q − M2 − M2T + hZ3 , Ψ25 = BK2 , Ψ55 = −h−1 R −1
Proof Choose a Lyapunov-Krasovskii functional as

t
V (xt ) =x T (t)P x(t) + x T (s)Qx(s)ds
t−h
0 t
+ dθ ẋ T (s)R ẋ(s)ds (6.9)
−h t+θ
where xt = x(t + s), s ∈ [−h, 0], P > 0, Q > 0, and R > 0.

First, we consider the robust stability of closed-loop system (6.5) with ζ (t) = 0.
Taking the derivative of V (xt ) with respect to t along the trajectory of system (6.5)
yields
V̇ (xt ) =x T (t)[P A(t) + AT (t)P + Q]x(t) + hẋ T (t)R ẋ(t)

+ 2x T (t)P B(t)x(t − h) + 2x T (t)P Df (x, t)
t
− x T (t − h)Qx(t − h) − ẋ T (s)R ẋ(s)ds (6.10)
t−h
By Lemma 2.7, for any 6 × 6 matrices R > 0, M1 , M2 , Z1 ≥ 0, Z2 , and Z3 ≥ 0

satisfying the constraint (6.7), we have
t
− ẋ T (s)R ẋ(s)ds ≤ x T (t)(M1T + M1 + hZ1 )x(t)
t−h
+ 2x T (t)(−M1T + M2 + hZ2 )x(t − h)

+ x T (t − h)(−M2T − M2 + hZ3 )x(t − h) (6.11)
T
Denote ξ(t) = x T (t) x T (t − h) f T (x, t) . Then, from Eqs. (6.10) and
(6.11), together with Eq. (2.43), we obtain
V̇ (xt ) ≤ ξ T (t)[∇(t) + Γ T (t)(hR)Γ (t)]ξ(t) (6.12)
where
⎧ ⎡ ⎤
⎪
⎪ ∇11 (t) ∇12 (t) P D
⎨ ∇(t) = ⎣ ∗ Ψ22 0 ⎦
(6.13)
⎪
⎪ ∗ ∗ −I
⎩
Γ (t) = Γ11 (t) Γ12 (t) D
with
⎧
⎪
⎪ ∇11 (t) = Ψ11 + μ2 I + P M̂ F̂ (t)N̂ + N̂ T F̂ T (t)M̂ T P
⎪
⎪
⎪
⎨ +P BV G(t)U1 + U1T GT (t)V T B T P
∇12 (t) = Ψ12 + P BV G(t)U2 (6.14)
⎪
⎪
⎪ Γ11 (t) = Ψ15 + M̂ F̂ (t)N̂ + BV G(t)U1
⎪
⎪
⎩
Γ12 (t) = Ψ25 + BV G(t)U2
Then, a sufficient condition for robust stability of system (6.5) with ζ (t) = 0 is that
there exist 6 × 6 matrices P > 0, Q > 0, R > 0, M1 , M2 , Z1 ≥ 0, Z2 , Z3 ≥ 0,
1 × 6 matrices K1 and K2 such that (6.7) and
∇(t) + Γ T (t)(hR)Γ (t) < 0 (6.15)

By Schur complement and S-procedure, matrix inequality (6.15) holds if there

exist scalars ε > 0 and ! > 0 such that
⎡ T ⎤
Ψ11 Ψ12 PD Ψ15 μI εP BV U1T !P M̂ N̂ T
⎢ ∗ Ψ22 0 T
Ψ25 0 0 U2T 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I DT 0 ⎥
⎢ 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ Ψ55 0 εBV 0 ! M̂ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ −I 0 0 0 0 ⎥<0 (6.16)
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
which can be derived from the matrix inequality (6.6). It indicates that if the matrix
inequalities (6.6) and (6.7) hold, then the closed-loop system (6.5) with ζ (t) = 0 is
robustly stable.
We now prove that the H∞ performance (6.1) is guaranteed for nonzero ζ (t)
and a prescribed γ > 0. Taking the derivative of V (xt ) with respect to t along the
trajectory of (6.5), and combing with (4.1), after simple manipulations, one yields
V̇ (xt )+ηT (t)η(t) − γ 2 ζ T (t)ζ (t)

¯
≤ ς T (t)[∇(t) + Υ T Υ + Γ¯ T (t)(hR)Γ¯ (t)]ς (t) (6.17)
where
⎧ T
⎪
⎪ ς (t) = x T (t) x T (t − h) f T (x, t) ζ T (t)
⎪
⎪ ⎡ ⎤
⎪
⎪ ∇11 (t) ∇12 (t) P D P D0
⎪
⎪
⎪
⎨¯ ⎢ ∗ Ψ22 0 0 ⎥
∇(t) = ⎢⎣ ∗
⎥
∗ −I 0 ⎦ (6.18)
⎪
⎪
⎪
⎪ ∗ ∗ ∗ −γ 2 I
⎪
⎪
⎪
⎪ Υ = C1 0 0 D 1
⎪
⎩
Γ¯ (t) = Γ11 (t) Γ12 (t) D D0
If the matrix inequalities (6.6) and (6.7) hold, by using Schur complement and
S-procedure again, one yields
¯
∇(t) + Υ T Υ + Γ¯ T (t)(hR)Γ¯ (t) < 0 (6.19)
which leads to
V̇ (xt ) + ηT (t)η(t) − γ 2 ζ T (t)ζ (t) < 0 (6.20)

Since V (x(0)) = 0 under zero initial condition, integrating both sides of Eq. (6.20)
from 0 to ∞, we have
∞
[ηT (t)η(t) − γ 2 ζ T (t)ζ (t)]dt < 0 (6.21)
0
which means that the H∞ performance (6.1) is guaranteed. This completes the
proof. Notice that there exist nonlinear terms in matrix inequality (6.6). We give an
equivalent form of Proposition 6.1, where the sufficient condition for the existence
of the delayed robust non-fragile H∞ control law is formulated in the form of a
linear matrix inequality.
Proposition 6.2 Under Assumption 1 of Chap. 4, for given scalars γ > 0, μ > 0,
and h > 0, the closed-loop system (6.5) with ζ (t) = 0 is robustly stable, and the
H∞ performance (6.1) is guaranteed, if there exist 6 × 6 matrices P̄ > 0, Q̄ > 0,
R̄ > 0, M̄1 , M̄2 , Z̄1 ≥ 0, Z̄2 , Z̄3 ≥ 0, 1 × 6 matrices K̄1 , K̄2 , scalars ε > 0 and
! > 0 such that
⎡ ⎤
Φ11 Φ12 D D0 Φ15 Φ16 μP̄ Φ18 Φ19 ! M̂ P̄ N̂ T
⎢ ∗ Φ 0 ⎥
⎢ 22 0 0 Φ25 0 0 0 Φ29 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I 0 hD T 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ −γ 2 I hD0T D1T 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ −hR̄ 0 0 Φ58 0 h! M̂ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ −I 0 0 0 ⎥
⎢ 0 0 ⎥<0 (6.22)
⎢ ∗ ∗ ∗ ∗ ∗ ∗ −I ⎥
⎢ 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
and
⎡ ⎤
2P̄ − R̄ M̄1 M̄2
⎣ ∗ Z̄1 Z̄2 ⎦ ≥ 0 (6.23)
∗ ∗ Z̄3
where
⎧
⎪
⎪ Φ11 = AP̄ + P̄ AT + B K̄1 + K̄1T B T + Q̄ + M̄1T + M̄1 + hZ̄1
⎪
⎪
⎪
⎨ Φ12 = B K̄2 − M̄1T + M̄2 + hZ̄2 , Φ15 = hP̄ AT + hK̄1T B T
Φ16 = P̄ C1T , Φ18 = εBV , Φ19 = P̄ U1T (6.24)
⎪
⎪
⎪
⎪ Φ22 = −Q̄ − M̄2 − M̄2 + hZ̄3 , Φ25 = hK̄2T B T
T
⎪
⎩Φ
29 = P̄ U2T , Φ58 = hεBV
Moreover, the gain matrices in Eq. (6.2) are given by Ki = K̄i P̄ −1 , i = 1, 2.

Proof Pre- and post-multiplying matrix inequality (6.6) by
diag{P −1 , P −1 , I, I, hI, I, I, I, I, I, I }
and its transpose, respectively, and setting
P̄ = P −1 , Q̄ = P −1 QP −1 , R̄ = R −1
K̄i = Ki P −1 , M̄i = P −1 Mi P −1 , i = 1, 2
Z̄j = P −1 Zj P −1 , j = 1, 2, 3
we obtain the linear matrix inequality (6.22).

On the other hand, the matrix inequality (6.7) is equivalent to the following
matrix inequality:
⎡ ⎤
P̄ R P̄ M̄1 M̄2
⎣ ∗ Z̄1 Z̄2 ⎦ ≥ 0 (6.25)
∗ ∗ Z̄3
Notice that P̄ R P̄ ≥ 2P̄ − R̄. It is clear that if the matrix inequality (6.23) holds,
then we have Eq. (6.25). This completes the proof.

As a special case of the control law (6.2), we present a delay-free robust non-
fragile H∞ control law as
u(t) = [K1 + ΔK1 (t)]x(t) (6.26)
In this case, the closed-loop system (6.5) can be written as
ẋ(t) = A(t)x(t) + Df (x, t) + D0 ζ (t), x(0) = x0 (6.27)
A sufficient condition for the existence of the control law (6.26) is given by the
following corollary, which can be obtained from Proposition 6.2.
Corollary 6.1 Under Assumption 1 of Chap. 4, for given scalars μ > 0 and
γ > 0, the closed-loop system (6.27) with ζ (t) = 0 is robustly stable, and the
H∞ performance (6.1) is guaranteed, if there exist a 6 × 6 matrix P̄ > 0, a 1 × 6
matrix K̄1 , and scalars ε > 0 and ! > 0 such that
⎡ ⎤
Σ D D0 P̄ C1T μ̄P εBV P̄ U1T ! M̂ P̄ N̂ T
⎢ ∗ −I 0 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I D T 0 ⎥
⎢ 1 0 0 0 0 ⎥
⎢∗ ∗ ∗ −I 0 0 ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 0 0 ⎥<0 (6.28)
⎢ ⎥
⎢∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥⎥
⎢
⎢∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
where
Σ = P̄ AT + AP̄ + K̄1T B T + B K̄1
Moreover, the gain matrix K1 in Eq. (6.26) is given by K1 = K̄1 P̄ −1 .

If the system’s parameter uncertainties are not considered, the system (2.47)
reduces to (2.50). Substituting Eq. (6.2) into (2.50) yields the closed-loop system
as
ẋ(t) =[A + BK1 + BΔK1 (t)]x(t)

+ B[K2 + ΔK2 (t)]x(t − h) + Df (x, t) + D0 ζ (t) (6.29)
In order to solve the gain matrices K1 and K2 of the delayed non-fragile H∞

control law (6.2), we present the following corollary, which can be derived from
Proposition 6.2.
Corollary 6.2 Under Assumption 1 of Chap. 4, for given scalars γ > 0, μ > 0,
and h > 0, the closed-loop system (6.29) with ζ (t) = 0 is asymptotically stable,
and the H∞ performance (6.1) is guaranteed, if there exist 6 × 6 matrices P̄ > 0,
Q̄ > 0, R̄ > 0, M̄1 , M̄2 , Z̄1 ≥ 0, Z̄2 , Z̄3 ≥ 0, 1 × 6 matrices K̄1 and K̄2 , and a
scalar ε > 0 such that (6.23) and
⎡ ⎤
Φ11 Φ12 D D0 Φ15 Φ16 μP̄ Φ18 Φ19
⎢ ∗ Φ22 0 0 Φ25 0 0 0 Φ29 ⎥
⎢ ⎥
⎢ ∗ ∗ −I 0 hD T 0 ⎥
⎢ 0 0 0 ⎥
⎢ ∗ ∗ ∗ −γ 2 I hD0T D1T 0 ⎥
⎢ 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ −hR̄ 0 0 Φ58 0 ⎥ < 0 (6.30)
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ −I 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI
Moreover, the gain matrices K1 and K2 in Eq. (6.2) are given by Ki = K̄i P̄ −1 ,
i = 1, 2.
Correspondingly, if a delay-free non-fragile H∞ control law in the form (6.26)
is utilized to control the nominal system (2.50), the resulting closed-loop system is
given as
ẋ(t) =[A + BK1 + BΔK1 (t)]x(t) + Df (x, t) + D0 ζ (t) (6.31)
To obtain the gain matrix K1 , we provide the following corollary.

Corollary 6.3 Under Assumption 1 of Chap. 4, for given scalars μ > 0 and γ > 0,
the closed-loop system (6.31) with ζ (t) = 0 is asymptotically stable, and the H∞
performance (6.1) is guaranteed, if there exist a 6 × 6 matrix P̄ > 0, a 1 × 6 matrix
K̄1 , and a scalar ε > 0 such that
⎡ ⎤
Σ D D0 P̄ C1T μP̄ εBV P̄ U1T
⎢∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥
⎢∗ ∗ −γ 2 I D1T ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ −I 0 0 0 ⎥<0 (6.32)
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
Moreover, the gain matrix in Eq. (6.26) is given by K1 = K̄1 P̄ −1 .

To compare with a nonlinear control scheme [65], a delayed dynamic output
feedback control scheme [100], and an integral sliding mode control scheme [99],
we consider a delayed state feedback control scheme, which is a special case of
the delayed non-fragile H∞ control scheme. Let ΔK1 (t) ≡ 0 and ΔK2 (t) ≡ 0 in
Eq. (6.2). Then, we obtain a delayed state feedback control law as
u(t) = K1 x(t) + K2 x(t − h) (6.33)
In this situation, substituting Eq. (6.33) into (2.45) yields
ẋ(t) = (A + BK1 )x(t) + BK2 x(t − h) + Df (x, t). (6.34)
By Corollary 6.2, we have the following result.

Corollary 6.4 For given scalars μ > 0 and h > 0, the closed-loop system (6.34) is
asymptotically stable, if there exist 6 × 6 matrices P̄ > 0, Q̄ > 0, R̄ > 0, M̄1 , M̄2 ,
Z̄1 ≥ 0, Z̄2 , Z̄3 ≥ 0, 1 × 6 matrices K̄1 and K̄2 such that (6.23) and
⎡ ⎤
Φ11 Φ12 D Φ15 μP̄
⎢ ∗ ⎥
⎢ Φ22 0 Φ25 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I hD T 0 ⎥<0 (6.35)
⎢ ⎥
⎣ ∗ ∗ ∗ −hR̄ 0 ⎦
∗ ∗ ∗ ∗ −I
Moreover, the gain matrices K1 and K2 in Eq. (6.33) are given by Ki = K̄i P̄ −1 ,
i = 1, 2.
6.3 Simulation Results and Discussions
In this section, an example is given to show the effectiveness of the delayed

non-fragile H∞ control scheme for the offshore steel jacket platform. First, the
performance of the nominal system with the delayed non-fragile H∞ control scheme
and with the delay-free non-fragile H∞ control scheme are investigated. Second, the
performance of the uncertain system with the delayed robust non-fragile H∞ control
scheme and with the delay-free one are compared from the controlled oscillation
amplitudes of the offshore platform and the required control force. Finally, the
delayed state feedback control scheme is compared with a nonlinear control scheme
[65], a delayed dynamic output feedback control scheme [100], and an integral
sliding mode control scheme [99], respectively.
The parameters of an offshore platform and the data of the wave are taken from
[65] and listed in Table 4.1 The matrices C1 and D1 in the output Eq. (4.1) are given
as

010000 0.1 0
C1 = , D1 = (6.36)
000100 0 0.1
The nonlinear self-excited wave force f (x, t) can be computed as [65]. The
external disturbance acting on the first and second modes are simulated by uniformly
distributed random signals with maximum amplitudes 4.6 × 105 and 1.1 × 105 N,
respectively.
6.3.1 Performance of the Nominal System
For comparison purpose, we first investigate the delay-free non-fragile H∞ control.

Let γ = 0.21, μ = 0.8. It is assumed that the parameters of perturbation in
controller gain are given by

V = 26 0 26 0 26 0 , U1 = 21, G(t) = 0.9sin(t).
Then, by Corollary 6.3, one can obtain a non-fragile H∞ controller (NFHC) of the
form (6.26), where the gain matrix K1 is given as

K1 = 105 × −1.1860 0.2374 2.7705 −0.0647 −0.4528 −0.9704 .
Under the obtained NFHC, the oscillation amplitudes of the first, second, and third
floors of the nominal system (2.50) are 0.2013, 0.2185, and 0.2318 m peak to peak,
respectively. The maximum control force required is about 3.2524×105 N. Depicted
in Fig. 6.1 are the responses of the first, second, and third floors of the offshore
platform and the required control force, respectively.
0.15 0.15
Second Floor (m)

0.1 0.1
First Floor (m)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) 5 (b)
x 10
0.15 3
Control Force (N)

0.1 2
Third Floor (m)
0.05
1
0
0
−0.05
−0.1 −1
−0.15 −2
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) (d)
Fig. 6.1 Responses of the system (2.50) and the control force under NFHC
Then, we focus on studying the delayed non-fragile H∞ control. Set h = 0.03 s

and U2 = 20. The values of other parameters μ, γ , V , U1 , and G(t) are the same
as those given above. By Corollary 6.2, we obtain the gain matrices K1 and K2 of a
delayed non-fragile H∞ controller (DNFHC) as

K1 = 105 × −0.5106 0.0359 2.0787 0.0154 −0.1661 −0.3978 ,

K2 = 102 × 0.7809 −0.0752 −1.0356 0.2059 0.2879 0.5191 .
When the DNFHC is applied to control the nominal system (2.50), the response of
the three floors and the control force are showed in Fig. 6.2. It can be seen that the
oscillation amplitudes of the first, second, and third floors peak to peak are 0.1989,
0.2168, and 0.2309 m, respectively, and the range of the control force peak to peak
is about 1.7288 × 105 N.
The oscillation amplitudes of the three floors and the required control force are
listed in Table 6.1, where the oscillation amplitudes of the three floors of the system
without control are presented [90]. It can be seen from Table 6.1 that under the
NFHC and the DNFHC, the average oscillation amplitudes of the three floors of the
system are reduced to 14.3% and 14.2% of that without control, respectively. The
reduction of the vibration of the three floors under the two non-fragile controllers
are almost at the same level. However, it is clear to see that the control force by
0.15 0.15
0.1 0.1
Second Floor (m)

First Floor (m)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) 5 (b)
x 10
0.15 3
0.1
Control Force (N)

2
Third Floor (m)
0.05
1
0
0
−0.05
−0.1 −1
−0.15 −2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 6.2 Responses of the system (2.50) and the control force under DNFHC
Table 6.1 Maximum oscillation amplitudes of the nominal system (2.50) and the range of the
control force under different controllers
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
No control 1.4159 1.5270 1.6061 –
NFHC 0.2013 0.2185 0.2318 325,240
DNFHC 0.1989 0.2168 0.2309 172,880
the NFHC is 1.88 times as that by the DNFHC. This indicates that by intentionally
introducing a time-delay into the control channel, the required control force can be
reduced.
To investigate the effects of introduced time-delays on the non-fragile H∞
control for the system, we apply DNFHC to control the nominal system, and let
time-delay h take different values. Then we calculate the peak to peak oscillation
amplitudes of the three floors and the range of the required control force, which are
listed in Table 6.2, from which it can be seen that:
• Under the DNFHC, as h ≤ 46.0 s, the average reduction of the vibration of three
floors of the nominal system is the almost same as the one under the delay-free
NFHC, while the required control force by the DNFHC is much smaller than the
one by the NFHC, which shows that time-delays properly introduced can make
a positive contribution to the non-fragile H∞ control for the offshore steel jacket
platform.
Table 6.2 Maximum oscillation amplitudes of the nominal system (2.50) and the range of the
control force under the delayed non-fragile H∞ controller (DNFHC) for different values of time-
delay h
h (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
0.05 0.1989 0.2169 0.2310 173,020
0.10 0.1990 0.2169 0.2310 172,930
1.00 0.1983 0.2164 0.2307 171,850
5.00 0.1977 0.2158 0.2301 170,870
15.0 0.1983 0.2164 0.2307 171,890
25.0 0.1988 0.2168 0.2311 172,580
35.0 0.1989 0.2168 0.2309 173,200
46.0 0.2003 0.2191 0.2339 198,630
47.0 0.2035 0.2217 0.2353 419,500
48.0 0.2339 0.2551 0.2720 484,630
• Under the DNFHC, the allowable time-delay can be obtained as 46.0 s. Com-
pared with the delayed dynamic output feedback controller (DDOFC) [100],
where the upper bound of the introduced time-delay with positive effects is less
than 0.11 s, the DNFHC can provide more options of the time-delay to improve
the control performance of the offshore platform.
6.3.2 Performance of the Uncertain System
It is assumed that the maximum perturbation ω̂i = 0.02, i = 1, 2, T , ξ̂1 = ξ̂2 =

0.005, and ξ̂T = 0.01. Then by (2.48), the matrices M̂ and N̂ in system (2.47) can
be obtained as (4.35).
Set γ = 0.58 and μ = 0.8. The time-varying perturbation matrix F̂ (t) in the
system is given by F̂ (t) = sin(t) · I , the perturbation parameters in the controller
gain are assumed to be

V = 12 0 12 0 12 0 , U1 = 7, G(t) = 0.9sin(t).
Then, by Corollary 6.1, we design a robust non-fragile H∞ controller (RNFHC) in

the form (6.26), where the gain matrix K1 is obtained as

K1 = 105 × −1.3460 0.4222 2.2046 −0.1235 −0.5328 −1.1850 .
Under the RNFHC, the responses of the three floors and the required control
force are showed in Fig. 6.3. It can be obtained that the peak-to-peak oscillation
amplitudes of the first, second, and third floors are 0.2068, 0.2237, and 0.2365 m,
respectively. The maximum control force is about 5.8260 × 105 N.
(a) (b)
0.15 0.15
0.1 0.1
Second Floor (m)

First Floor (m)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) 5 (d)
x 10
0.15 4
0.1
Control Force (N)

Third Floor (m)
2
0.05
0 0
−0.05
−2
−0.1
−0.15 −4
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 6.3 Responses of the system (2.47) and the control force under RNFHC
Now, let h = 0.03 s and U2 = 12I6 . Then we design a delayed robust non-fragile
H∞ controller (DRNFHC) by Proposition 6.2, and the gain matrices K1 and K2 can
be computed as

K1 = 105 × −1.4732 0.3890 3.1569 −0.1310 −0.5498 −1.2545 ,

K2 = 102 × −4.9699 1.5278 3.4637 −0.8936 −2.1829 −4.6154 .
As the DRNFHC is used to control the uncertain system, the oscillation

amplitudes of the first, second, and third floors of the uncertain system (2.47) are
0.2059, 0.2235, and 0.2363 m, respectively. The maximum control force required is
4.9713×105 N. The responses of the three floors of the system and the control force
are given by Fig. 6.4.
To compare two robust non-fragile H∞ control schemes, the control force and the
oscillation amplitudes of the three floors of the uncertain system under no control
[90], RNFHC and DRNFHC are listed in Table 6.3. It can be seen that under
RNFHC and DRNFHC, the average oscillation amplitudes of the three floors of
the uncertain system are both reduced to about 14.5% of the one when no controller
is applied. However, the control force required by RNFHC is 1.17 times as that by
DRNFHC, which indicates that to obtain the same control performance, the control
force required by DRNFHC is smaller than that by RNFHC. In fact, as demonstrated
in Table 8.4, DRNFHC can achieve the same control performance as RNFHC, while
it takes less control cost than RNFHC as time-delay h ≤ 46.0 s (Table 6.4).
0.15 0.15
Second Floor (m)

0.1 0.1
First Floor (m)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) x 10
5 (b)
4
Control Force (N)

0.1
Third Floor (m)
2
0.05
0 0
−0.05
−2
−0.1
−4
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) (d)
Fig. 6.4 Responses of the system (2.47) and the control force under DRNFHC
Table 6.3 Maximum oscillation amplitudes of the uncertain system (2.47) and the range of the
control force under different controllers
Controllers Floor 1(m) Floor 2(m) Floor 3(m) Control force(N)
No control 1.4180 1.5371 1.6181 –
RNFHC 0.2068 0.2237 0.2365 582,600
DRNFHC 0.2059 0.2235 0.2363 497,130
6.3.3 Comparison of Several Controllers
Let h = 0.2 s and μ = 0.8. Then, by Corollary 6.4, a delayed state feedback
controller (DSFC) in the form (6.33) is obtained, where

K1 = 104 × −0.2080 −0.0631 1.9832 0.4715 0.0738 −0.1780 ,

K2 = 7.3945 −0.6173 −37.3389 0.9779 3.0632 5.8249 .
When the obtained DSFC is applied to control the system (2.45), the responses
of the three floors of the system and the control force are presented in Fig. 6.5. It can
be computed that the oscillation amplitudes of the first, second, and third floors of
the system are reduced to 0.1873, 0.2054, and 0.2197 m, respectively, and the range
of the control force required is about 7.3460 × 104 N. Table 6.5 lists the range of the
control force and the oscillation amplitudes of the three floors of the system (2.45)
Table 6.4 Maximum oscillation amplitudes of the uncertain system (2.47) and the control force
under the delayed robust non-fragile H∞ controller (DRNFHC) for different values of time-delay h
h (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
0.05 0.2059 0.2235 0.2363 497,720
0.10 0.2059 0.2235 0.2363 497,140
0.50 0.2059 0.2235 0.2364 498,000
1.00 0.2059 0.2235 0.2364 497,310
5.00 0.2059 0.2234 0.2363 497,330
10.0 0.2059 0.2235 0.2363 497,780
15.0 0.2059 0.2235 0.2364 498,100
20.0 0.2058 0.2234 0.2363 497,630
30.0 0.2066 0.2234 0.2363 496,810
35.0 0.2059 0.2235 0.2364 498,270
40.0 0.2075 0.2235 0.2363 497,520
46.0 0.2059 0.2235 0.2364 559,480
47.0 0.2097 0.2261 0.2370 687,110
48.0 0.2058 0.2234 0.2363 675,460
49.0 0.3845 0.4269 0.4639 887,120
0.15 0.15
Second Floor (m)
0.1 0.1
First Floor (m)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) 5 (b)
x 10
0.15 1.5
Control Force (N)
0.1 1
Third Floor (m)
0.05
0.5
0
0
−0.05
−0.1 −0.5
−0.15 −1
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) (d)
Fig. 6.5 Responses of the system (2.45) and the control force under DSFC
under the different controllers: nonlinear controller (NLC) [65], DDOFC [100], and
integral sliding mode controller (ISMC) [99]. From Table 6.5, one observes that:
6.5 Notes 107
Table 6.5 Maximum oscillation amplitudes of the system (2.45) and the range of the control
force with different controllers
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
NLC [65] 0.3050 0.3050 0.3050 200,000
DDOFC [100] 0.2329 0.2543 0.2705 400,000
SMC [99] 0.2192 0.2301 0.2383 215,700
DSFC 0.1873 0.2054 0.2197 73,460
• The oscillation amplitudes of the three floors of the offshore platform under
DSMC are much smaller than the ones under NLC, DDOFC, and ISMC.
Moreover, the range of the control force required by DSMC is much less than
the one by NLC, DDOFC and ISMC.
• It can be computed that under the obtained DSFC, the maximum allowable time-
delay with positive effects on the control performance of the system (2.45) can
reach 60 s, which is much larger than the one under DDOFC.
6.4 Conclusions
This chapter has developed a delayed non-fragile H∞ control scheme for the
offshore platform and investigated the positive effects of time-delays on the non-
fragile H∞ control for the system. The simulation results have shown that by
properly introduced time-delay into control channel, the control force by the delayed
non-fragile H∞ control scheme can be significantly reduced. Moreover, the obtained
allowable maximum time-delay with the delayed non-fragile H∞ control scheme
has been much larger than the one with delayed dynamic output feedback control
scheme.
6.5 Notes
The main results of this chapter are taken from [91], where the non-fragile H∞
control scheme is utilized to deal with perturbation of controller gain and external
disturbance, and both current and delayed states are used to design the active
controller for an offshore platform with TMD mechanisms. In Zhang, Han, and
Huang [108], a pure delayed non-fragile controller is proposed to stabilize the
offshore platform subject to self-excited wave force, and the positive effects of
time-delays on the performance of the system are further investigated. It should be
pointed out that in [108] only pure delayed sates are used as feedback signals. Such
an idea is also applied to an offshore steel jacket platform with AMD mechanisms,
and a pure delayed H∞ controller is designed to attenuate the vibration induced by
external wave force. As for the detailed discussion of the controller, one refers to
Zhang and Tang [89].
Notice that intensionally introducing proper time-delays can improve the control
performance of offshore platform systems. However, some challenging issues need
to be considered. For example, delayed feedback control schemes are mainly based
on simplified dynamic models of offshore steel jacket platforms, where only the
dominant vibration modes are considered, while other higher vibration modes are
ignored. To develop general dynamic models for offshore platforms and then to
design effective delayed feedback control schemes are still interesting topics for
active control of offshore platforms. In addition, the obtained results are mainly
based on theoretical analysis and simulations. From the implementation point of
view, how to utilize these theoretical results in practical offshore platforms should
be reevaluated.
Chapter 7
Delayed Dynamic Output Feedback
Control
This chapter investigates the effect of a time-delay on dynamic output feedback

control of an offshore steel jacket platform subject to a nonlinear wave-induced
force. First, a conventional dynamic output feedback controller is designed to reduce
the internal oscillations of the offshore platform. It is found that the designed
controller is of a larger gain in the sense of Euclidean norm, which demands a
larger control force. Second, a time-delay is introduced intentionally to design a
new dynamic output feedback controller such that (i) the controller is of a small
gain in the sense of Euclidean norm and (ii) the internal oscillations of the offshore
platform can be dramatically reduced. It is shown through simulation results that
purposefully introducing time-delays can be used to improve control performance.
This chapter is organized as follows. A dynamic output feedback control scheme
is presented for an offshore steel jacket platform with the TMD mechanisms
in Sect. 7.1. By artificially introducing a time-delay into the control channel, a
delayed dynamic output feedback control scheme is developed in Sect. 7.2, where
an iterative algorithm is proposed to solve the delayed output feedback controller. In
Sect. 7.3, based on simulation results, the comparison between the dynamic output
feedback control scheme and the delayed dynamic output feedback control scheme
is made, and the effects of the purposefully introduced time-delays on the dynamic
output feedback control for the offshore steel jacket platform are investigated.
Section 7.4 concludes the chapter, and Sect. 7.5 provides a brief note.
7.1 Dynamic Output Feedback Control
Consider a simple offshore steel jacket platform in Fig. 2.2 [65, 72], the dynamic
model of the offshore platform is given by (2.45).

https://doi.org/10.1007/978-981-13-2986-9_7
110 7 Delayed Dynamic Output Feedback Control
Let y ∈ Rr denote the output vector of the system
y(t) = Cx(t) (7.1)
where C ∈ Rr×6 is a constant matrix. The corresponding state space model of the
offshore platform can be expressed as
⎧
⎪
⎪ ẋ(t) = Ax(t) + Bu(t) + Df (x, t)
⎨
y(t) = Cx(t) (7.2)
⎪
⎪
⎩
x(t0 ) = x0 , t0 ≥ 0
We seek a dynamic output feedback controller of the form

ẋc (t) = AK xc (t) + BK y(t)
(7.3)
u(t) = CK xc (t) + DK y(t)
where xc ∈ R6 and AK ∈ R6×6 , BK ∈ R6×r , CK ∈ R1×6 , and DK ∈ R1×r

are real matrices to be determined, such that the resulting closed-loop system is
asymptotically stable.
Introduce an augmented vector
ϑ(t) := [x T (t) xcT (t)]T
Then the resulting closed-loop system by (7.2) and (7.3) is given by
ϑ̇(t) = (A0 + H KL)ϑ(t) + E T Df (x, t) (7.4)
where E = [I 0], A0 = diag{A, 0}, H = diag{B, I }, L = diag{C, I } with

I ∈ R6×6 and 0 ∈ R6×6 , and

D K CK
K := (7.5)
BK AK
For system (7.4), we have the following result.

Proposition 7.1 Let W1 and W2 be the orthogonal complements of B and C T ,
respectively. A dynamic output feedback controller of form (7.3) is solvable for
system (7.2) if there exist 6 × 6 matrices X > 0 and Y > 0 such that
T
W1 (AX + XAT + DD T )W1 μW1T X
<0 (7.6)
∗ −I
T
W2 (Y A + AT Y + μ2 I )W2 W2T Y D
<0 (7.7)
∗ −I

X I
>0 (7.8)
∗ Y
7.1 Dynamic Output Feedback Control 111
Moreover, if (7.6), (7.7) and (7.8) are feasible on matrix variables X and Y , then the
controller parameters K defined in (7.5) can be obtained by solving the following
LMI:
P (A0 + H KL) + (A0 + H KL)T P + μ2 E T E + P E T DD T EP < 0 (7.9)
where
−1
y I I X
P := , MN T = I − XY (7.10)
NT 0 0 MT
Proof Choose a Lyapunov function for system (7.4)
V (ϑ(t)) = ϑ T (t)P ϑ(t) (7.11)
where P = P T > 0 is to be determined. Then taking the derivative of V (ϑ(t)) with

respect to t along the trajectory of system (7.4) yields
V̇ (ϑ(t)) = 2ϑ T (t)P ϑ̇(t)

= 2ϑ T (t)P (A0 + H KL)ϑ(t) + 2ϑ T (t)P E T Df (x, t) (7.12)
Noting that (2.43), we have
f T (x, t)f (x, t) ≤ μ2 x T (t)x(t)
Since x(t) = Eϑ(t), one gets
f T (x, t)f (x, t) ≤ μ2 ϑ T (t)E T Eϑ(t)
which leads to
0 ≤ μ2 ϑ T (t)E T Eϑ(t) − f T (x, t)f (x, t) (7.13)
From (7.12) and (7.13), it is clear that
V̇ (ϑ(t)) ≤ 2ϑ T (t)P (A0 + H KL)ϑ(t) + 2ϑ T (t)P E T Df (x, t)

+ μ2 ϑ T (t)E T Eϑ(t) − f T (x, t)f (x, t)
:= ξ T (t)Φξ(t)
where ξ(t) := [ϑ T (t) f T (x, t)]T and

P (A0 + H KL) + (A0 + H KL)T P + μ2 E T E P E T D
Φ :=
∗ −I
Clearly, if Φ < 0, then V̇ (ϑ(t)) ≤ −λmin (−Φ)ξ T (t)ξ(t) ≤ −λmin (−Φ)ϑ T (t)
ϑ(t) < 0 for ϑ(t) = 0, which guarantees the asymptotic stability of system (7.4).
In order to solve the controller parameters, we rewrite Φ < 0 as
Φ = Φ0 + ΣΠ KΛT + (ΣΠ KΛT )T < 0 (7.14)
where

P A0 + AT0 P + μ2 E T E P E T D
Φ0 :=
∗ −I
T
P 0 H L
Σ := , Π := , Λ := .
0 I 0 0
By Lemma 2.4, the inequality (7.14) is feasible for the matrix variable K if and only
if
Π⊥T Σ −1 Φ0 Σ −1 Π⊥ < 0 (7.15)

ΛT⊥ Φ0 Λ⊥ < 0 (7.16)
To simplify the left sides of the above inequalities, let

Y N −1 X M
P = , P = (7.17)
NT % MT %
where M, N ∈ R6×6 and the symbol “%” denotes an irrelevant matrix. Choosing
⎡ ⎤ ⎡ ⎤
W1 0 W2 0
Π⊥ = ⎣ 0 0⎦ , Λ⊥ = ⎣ 0 0⎦
0 I 0 I
we have
−1
−1 −1 P 0 P A0 + AT0 P + μ2 E T E P E T D P −1 0
Σ Φ0 Σ =
0 I ∗ −I 0 I

A0 P −1 + P −1 AT0 + μ2 P −1 E T EP −1 E T D
=
∗ −I
7.1 Dynamic Output Feedback Control 113
⎡ ⎤
AX + XAT + μ2 XX AM + μ2 XM D
=⎣ ∗ μ2 M T M 0 ⎦
∗ ∗ −I
Thus, together with Schur complement, we have

T
W1 (AX + XAT + μ2 XX)W1 W1T D
Π⊥T Σ −1 Φ0 Σ −1 Π⊥ = <0
∗ −I
⇐⇒ W1T (AX + XAT + μ2 XX)W1 + W1T DD T W1 < 0
⇐⇒ W1T (AX + XAT + DD T )W1 + μ2 W1T XXW1 < 0
T
W1 (AX + XAT + DD T )W1 μW1T X
⇐⇒ <0
∗ −I
which is the LMI (7.6). Notice that

⎡ ⎤
Y A + AT y + μ2 I AT N Y D
ΛT⊥ Φ0 Λ⊥ = ΛT⊥ ⎣ ∗ 0 N T D ⎦ Λ⊥
∗ ∗ −I
T
W2 (Y A + AT y + μ2 I )W2 W2T Y D
=
∗ −I
It is clear that the inequality (7.16) is equivalent to LMI (7.7). On the other hand,
by Lemma 2.5, there exists a P > 0 satisfying (7.17) if and only if x − y −1 ≥ 0,
which is equivalent to (7.8). This completes the proof.

Now, based on Proposition 7.1, we design a dynamic output feedback con-
troller for system (2.45) by using the output information of the generalized
coordinates of vibrational modes 1 and 2 and the horizontal displacement of the
TMD. Let
⎡ ⎤
100000
C = ⎣0 0 1 0 0 0⎦ . (7.18)
000010
The parameters of the offshore platform system (2.45) are given as Table 4.1.
Applying Proposition 7.1, for μ = 0.8, the obtained dynamic output feed-
back controller, which is denoted by DOFC1, is given in (7.3) with AK , BK ,
CK , DK as
⎡ ⎤
−0.1245 0.0650 0.0148 0.0058 0.0226 −0.0013
⎢−0.0777 −0.4729 −0.0385 0.1210 0.1478 −0.0034⎥
⎢ ⎥
⎢ ⎥
⎢ 0.7365 1.1961 0.3529 −0.4959 −0.4414 0.0076 ⎥
AK := 10 × ⎢
5
⎥
⎢ 4.9878 0.5868 2.6570 −1.9726 −0.6940 −0.0126⎥
⎢ ⎥
⎣−3.6257 0.1269 −1.8758 0.9852 −0.1391 0.0202 ⎦
−5.7661 −2.5562 −0.0695 3.8665 5.9187 −0.1423
⎡ ⎤
0.0004 −0.0119 0.0023
⎢−0.0652 −0.0085 0.0231 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0.1581 −0.1373 −0.0748 ⎥
BK := 106 × ⎢ ⎥
⎢−0.0215 −1.6387 −0.1712⎥
⎢ ⎥
⎣ 0.1560 1.2877 0.0349 ⎦
−1.1523 −1.0792 0.8028

CK := 108 × −3.3396 2.7054 0.4868 −0.0658 0.3546 −0.0310

DK := 108 × 1.3256 −3.1587 0.2229
Under DOFC1, the responses of the first, the second, and the third floors are
depicted in Figs. 7.1, 7.2 and 7.3, from which one can see that the controlled
responses of the first, the second, and the third floors with oscillation magnitudes
of peak to peak have been reduced from 1.3793, 1.4946, and 1.5471 m to 0.2329,
0.2543, and 0.2705 m, respectively.
Notice that the Euclidean norm of the gain matrix A K BK
CK DK of the DOFC1 is
calculated as 5.5340 × 108 . One can see that the DOFC1 is of a large gain in the
sense of Euclidean norm, which demands a large control force.
0.15
0.1
0.05
FirstFloor Response (m)
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.1 Response of the first floor under the control of DOFC1
7.2 Design of a Delayed Dynamic Output Feedback Controller 115
0.15
0.1
SeondFloor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.2 Response of the second floor under the control of DOFC1
0.15
0.1
0.05
ThirdFloor Response (m)
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.3 Response of the third floor under the control of DOFC1
7.2 Design of a Delayed Dynamic Output Feedback

Controller
In practical feedback control systems, time-delays in the control action are

inevitable because of involved dynamics of actuators and sensors. As is seen
from the previous section, the obtained dynamic output feedback controller is of
a large gain in the sense of Euclidean norm. In this section, we are interested in
investigating the effect of a time-delay on dynamic output feedback control (DOFC)

of the offshore platform. More specifically, we introduce a time-delay h when we
measure the output of the system, i.e.
y(t) = Cx(t − h) (7.19)
where h > 0 is a parameter to be determined. The corresponding state space model

of the offshore platform is expressed as
⎧
⎪
⎪ ẋ(t) = Ax(t) + Bu(t) + Df (x, t)
⎨
y(t) = Cx(t − h) (7.20)
⎪
⎪
⎩
x(θ ) = φ(θ ), θ ∈ [t0 − h, t0 ].
where φ(θ ) is an initial function.
7.2.1 Delayed Dynamic Output Feedback Controller Design
We now design a dynamic output feedback controller of form (7.3) based on the
system (7.20). The resulting closed-loop system is
ϑ̇(t) = (A0 + H KL1 )ϑ(t) + H KL2 Eϑ(t − h) + E T Df (x, t) (7.21)
where A0 , H and K are the same as those in (7.4) and

00 C
L1 := , L2 :=
0I 0
We now state and establish the following stability criterion.

Proposition 7.2 For given positive scalars μ and h, the system (7.21) is asymp-
totically stable if there exist matrices P > 0, Q > 0 and R > 0 of appropriate
dimensions such that
⎡ ⎤
Θ11 P H KL2 + E T R P E T D h(A0 + H KL1 )T E T R
⎢ ∗ −Q − R 0 h(H KL2 )T E T R ⎥
Θ := ⎢
⎣ ∗
⎥<0
⎦ (7.22)
∗ −I hD T R
∗ ∗ ∗ −R
where
Θ11 := P (A0 + H KL1 ) + (A0 + H KL1 )T P + E T (Q − R + μ2 I )E. (7.23)

Proof Choose a Lyapunov-Krasovskii functional candidate as

t
V (t, ϑt ) = ϑ T (t)P ϑ(t) + ϑ T (s)E T QEϑ(s)ds
t−h
0 t
+h ds ϑ̇ T (θ )E T RE ϑ̇(θ )dθ (7.24)
−h t+s
where ϑt = ϑ(t + α), α ∈ [−h, 0] and P > 0, Q > 0, R > 0 to be determined.

Taking the derivative of V (t, ϑt ) with respect to t along the trajectory of (7.21)
yields
V̇ (t, ϑt ) = 2ϑ T (t)P ϑ̇(t) + ϑ T (t)E T QEϑ(t) − ϑ T (t − h)E T QEϑ(t − h)

t
+ h2 ϑ̇ T (t)E T RE ϑ̇(t) − h ϑ̇ T (θ )E T RE ϑ̇(θ )dθ
t−h
= ϑ (t)[P (A0 + H KL1 ) + (A0 + H KL1 )T P + E T QE]ϑ(t)

T
+ 2ϑ T (t)P H KL2 Eϑ(t − h) − ϑ T (t − h)E T QEϑ(t − h)

+ h2 [E(A0 + H KL1 )ϑ(t) + EH KL2 Eϑ(t − h) + Df (x, t)]T
× R[E(A0 + H KL1 )ϑ(t) + EH KL2 Eϑ(t − h) + Df (x, t)]
t
+ 2ϑ T (t)P E T Df (x, t) − h ϑ̇ T (θ )E T RE ϑ̇(θ )dθ (7.25)
t−h
Let
η(t) := [ϑ T (t) (Eϑ(t − h))T f T (x, t)]T

Γ := E(A0 + H KL1 ) EH KL2 D
Then
t
V̇ (t, ϑt ) = ηT (t)[Ψ1 + h2 Γ T RΓ ]η(t) − h ϑ̇ T (θ )E T RE ϑ̇(θ )dθ (7.26)
t−h
where
⎡ ⎤
ψ11 P H KL2 P E T D
Ψ1 := ⎣ ∗ −Q 0 ⎦
∗ ∗ 0
with
ψ11 := P (A0 + H KL1 ) + (A0 + H KL1 )T P + E T QE

For the integral term in (7.26), use Lemma 2.6 to obtain

t
−h ϑ̇ T (θ )E T RE ϑ̇(θ )dθ
t−h
T
Eϑ(t) −R R Eϑ(t)
≤
Eϑ(t − h) ∗ −R Eϑ(t − h)
T
ϑ(t) −E T RE E T R ϑ(t)
= (7.27)
Eϑ(t − h) ∗ −R Eϑ(t − h)
Notice that (2.43)
f T (x, t)f (x, t) ≤ μ2 x T (t)x(t) = μ2 ϑ T (t)E T Eϑ(t)
Then one has
0 ≤ μ2 ϑ T (t)E T Eϑ(t) − f T (x, t)f (x, t) (7.28)
From (7.28) and (7.27), it is clear that

t
−h ϑ̇ T (θ )E T RE ϑ̇(θ )dθ ≤ ηT (t)Ψ2 η(t) (7.29)
t−h
where
⎡ 2 T ⎤
μ E E − E T RE E T R 0
Ψ2 = ⎣ ∗ −R 0 ⎦
∗ ∗ −I
Substituting (7.29) into (7.26), we have

% &
V̇ (t, ϑt ) ≤ ηT (t) Ψ0 + h2 Γ T RΓ η(t) (7.30)
where Ψ0 := Ψ1 + Ψ2 , i.e.
⎡ ⎤
Θ11 P H KL2 + E T R P E T D
Ψ0 = ⎣ ∗ −Q − R 0 ⎦
∗ ∗ −I
with Θ11 being defined in (7.23).

On the other hand, if inequality (7.22) is feasible, then by Lemma 2.1, we have
Ψ0 + h2 Γ T RΓ < 0 (7.31)
So there exists a scalar δ = λmin (−(Ψ0 + h2 Γ T RΓ )) > 0 such that V̇ (t, ϑt ) ≤

−δηT (t)η(t) ≤ −δϑ T (t)ϑ(t) < 0 for ϑ(t) = 0, which guarantees the asymptotic
stability of the closed-loop system (7.21). This completes the proof.

Proposition 7.2 provides a delay-dependent stability criterion for system (7.21).
However, this condition cannot be used to design the controller parameters directly
due to some nonlinear terms, such as P H KL1 , P H KL2 and so on in matrix
inequality (7.22). In what follows, we propose a controller design method based
on Proposition 7.2. Similar to the proof of Proposition 7.1, a sufficient condition of
the existence of a desired controller can be obtained, which is stated in the following
proposition.
Proposition 7.3 Let W1 and [W2T W3T ]T be the orthogonal complements of C T
and [B T B T ]T , respectively. Matrix inequality (7.22) is feasible on matrix variable
K if and only if there exist 6 × 6 matrices X > 0, Y > 0, Q > 0 and R > 0 such
that (7.8) and

Ω11 W2T XR
Ω := <0 (7.32)
∗ −Q − R
⎡ ⎤
Ξ11 RW1 Y D hAT R
⎢ ∗ W T (−Q − R)W1 0 0 ⎥
⎢ 1 ⎥<0 (7.33)
⎣ ∗ ∗ −I hD T R ⎦
∗ ∗ ∗ −R
where
Ω11 := W2T [AX + XAT + X(Q − R + μ2 I )X]W2

+ (W2 + W3 )T DD T (W2 + W3 ) + W2T XAT W3
+ W3T AXW2 − h−2 W3T R −1 W3
Ξ11 := Y A + AT Y + Q − R + μ2 I.
One can see that matrix inequality (7.32) is still nonlinear on matrix variables,
which is a non-convex feasible problem. Now, we will convert this non-convex
feasible problem into a nonlinear minimization problem subject to a set of LMIs.
Define J := diag{I, X}. Then pre- and post-multiplying both sides of Ω in (7.32)
by J T and its transpose, respectively, yields

Ω11 W2T XRX
J T ΩJ = <0 (7.34)
∗ −XQX − XRX
Introducing two new matrix variables S > 0 and Z > 0 such that
XRX ≥ S, XQX ≥ Z
which are equivalent to, respectively,

R X−1 Q X−1
≥ 0, ≥ 0.
X−1 S −1 X−1 Z −1
Notice that
T
−W2T XRXW2 W2T XRX W2
=− XRX W2 −I
∗ −XRX −I
T
W2
≤− S W2 −I
−I

−W2T SW2 W2T S
= .
∗ −S
One can see that matrix inequality (7.34) is implied by

Ω̃11 W2T S
<0 (7.35)
∗ −Z − S
where
Ω̃11 := W2T [AX + XAT + X(Q + μ2 I )X − S]W2

+ (W2 + W3 )T DD T (W2 + W3 ) + W2T XAT W3
+ W3T AXW2 − h−2 W3T R −1 W3
By using Schur complement, (7.35) is equivalent to

⎡ ⎤
Ω̆11 W2T S W2T X μW2T X
⎢ ∗ −Z − S 0 0 ⎥
⎢ ⎥<0 (7.36)
⎣ ∗ ∗ −Q −1 0 ⎦
∗ ∗ ∗ −I
where
Ω̆11 :=W2T [AX + XAT − S]W2 − h−2 W3T R −1 W3

+ (W2 + W3 )T DD T (W2 + W3 )
+ W2T XAT W3 + W3T AXW2
Therefore, setting R̄ = R −1 , X̄ = X−1 , S̄ = S −1 , Z̄ = Z −1 , Q̄ = Q−1 and

similar to the proof of Proposition 7.2, we can derive a new sufficient condition
for the existence of the dynamic output feedback controller, which is stated in the
following result.
Proposition 7.4 Let W1 and [W2T W3T ]T be the orthogonal complements of C T and
[B T B T ]T , respectively. For given scalars μ > 0 and h > 0, the dynamic output
feedback control problem for system (7.20) is solvable if there exist 6 × 6 matrices
X > 0, Y > 0, Q > 0, R > 0, R̄ > 0, X̄ > 0, S > 0, S̄ > 0, Z > 0, Z̄ > 0,
and Q̄ > 0 such that (7.8), (7.33) and
⎡ ⎤
Υ W T S (W2T + W3T )D W2T X μW2T X
⎢ ∗ −Z2− S 0 ⎥
⎢ 0 0 ⎥
⎢ ⎥
⎢∗ ∗ −I 0 0 ⎥<0 (7.37)
⎢ ⎥
⎣∗ ∗ ∗ −Q̄ 0 ⎦
∗ ∗ ∗ ∗ −I

R X̄ Q X̄
≥ 0, ≥0 (7.38)
X̄ S̄ X̄ Z̄
R R̄ = I, XX̄ = I, S S̄ = I, Z Z̄ = I, QQ̄ = I (7.39)
where
Υ := W2T (AX + XAT − S)W2 + W2T XAT W3 + W3T AXW2 − h−2 W3T R̄W3 .
7.2.2 A Computational Algorithm
Proposition 7.4 is based on a set of LMIs subject to equality constraints. By

employing the cone complementary method proposed in [137], it can be converted
into a nonlinear minimization problem (NMP) subject to LMIs, which is stated in
the following.
Nonlinear Minimization Problem (NMP)
¯
Minimize Tr(X̄X + R̄R + Q̄Q + Z̄Z + SS) (7.40)
Subject to (7.8), (7.33), (7.37), (7.38) and

R I X I Q I
≥ 0, ≥ 0, ≥ 0, (7.41)
I R̄ I X̄ I Q̄

Z I SI
≥ 0, ≥ 0. (7.42)
I Z̄ I S̄
The following iterative algorithm can be used to solve the above NMP.
Algorithm 7.1 Solve the NMP (7.40)

Step 1 Find a feasible set Δ0 satisfying (7.8), (7.33), (7.37), (7.38) and
(7.41), (7.42), where Δ0 := (X0 , Y 0 , Q0 , R 0 , S 0 , Z 0 , X̄0 , Q̄0 , R̄ 0 , S̄ 0 , Z̄ 0 ).
Set l = 0.
Step 2 Solve the following LMI problem for the matrix variables Δ with Δ :=
(X, Y, Q, R, S, Z, X̄, Q̄, R̄, S̄, Z̄):
l
X̄ X + Xl X̄ + R̄ l R + R l R̄ + Q̄l Q
Minimize Tr
+Ql Q̄ + Z̄ l Z + Z l Z̄ + S̄ l S + S l S̄
Subject to (7.8), (7.33), (7.37), (7.38) and (7.41), (7.42)
Set Xl+1 = X, Ql+1 = Q, R l+1 = R, Z l+1 = Z, S l+1 = S, X̄l+1 = X̄,

Q̄l+1 = Q̄, R̄ l+1 = R̄, Z̄ l+1 = Z̄, S̄ l+1 = S̄.
Step 3 If matrix inequality (7.32) and
$ l $
$ $
$Tr X̄ x + x X̄ + R̄ R + R R̄ + Q̄ Q − 60$ < ε
l l l l
$ $ (7.43)
+Q Q̄ + Z̄ Z + Z Z̄ + S̄ S + S S̄
l l l l l
where ε is a prescribed sufficiently small positive scalar, are satisfied, then set
l = l + 1, and go to Step 2. If one of the conditions (7.32) and (7.43) is not
satisfied within a specified number of iterations, then exit.
Finally, if the nonlinear minimization problem (7.40) is feasible on the matrix
variables R, Q, and so on, then the desired dynamic output feedback controller of
form (7.3) can be obtained by solving the LMI (7.22) on the matrix variable K in
(7.5) with the known R, Q, and P of form (7.10).
7.2.3 Simulation Results
Now, we are in a position to design a dynamic output feedback controller for

system (2.45) based on Proposition 7.4 incorporating with Algorithm 7.1. Setting
h = 0.02, for μ = 0.8, the dynamic output feedback controller, which is denoted
by DOFC2, can be derived, and the controller parameters AK , BK , CK , DK are
given as
⎡ ⎤
−25.1098 2.4480 4.5903 −5.2317 −0.2174 −0.4551
⎢ 2.6316 −1.2776 −1.3518 0.1339 ⎥
⎢ 1.5660 0.0651 ⎥
⎢ ⎥
⎢ 11.0619 −9.9046 −16.8461 12.7973 −0.3222 −0.3411 ⎥
AK = ⎢ ⎥
⎢ 12.5650 −8.0020 −12.8012 5.1916 −0.4251 −0.8745 ⎥
⎢ ⎥
⎣ 33.7814 32.0469 33.7832 −4.2645 0.9524 −2.2384 ⎦
233.7041 71.3597 21.1233 36.5321 9.2835 −15.5076
7.3 Comparison Between Different Controllers 123
⎤
⎡
0.0075 −0.0002 0.0007
⎢ 0.0364 0.0005 0.0032 ⎥
⎢ ⎥
⎢ ⎥
⎢−23.4249 −43.5381 0.3757 ⎥
BK = ⎢ ⎥
⎢−20.5615 −43.6538 0.6321 ⎥
⎢ ⎥
⎣ 76.6976 37.5001 4.6865 ⎦
131.2817 −384.3427 33.3547

CK = 104 × −7.2556 0.4957 1.3130 −1.4967 −0.0622 −0.1302

DK = 2.8932 −0.0098 0.2151
Under DOFC2, the responses of the first, the second, and the third floors are
shown in Figs. 7.4, 7.5 and 7.6, from which one can see that the controlled responses
of the first, the second, and the third floors with oscillation magnitudes of peak to
peak have been reduced from 1.3793, 1.4946, and 1.5471 m to 0.2034, 0.2232, and
0.2391 m, respectively.
7.3 Comparison Between Different Controllers
In this section, based on the simulation results in Sects. 7.2 and 7.3, we make a
comparison between DOFC1 and DOFC2. Then, we investigate the effect of a time-
delay on the dynamic output feedback control of the offshore platform.
Firstly, we compare controller gains of controllers DOFC1 and DOFC2 in the
sense of the Euclidean norm. We calculate the Euclidean norms of the gain matrix
0.15
0.1
0.05
−0.05
−0.1
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.4 Response of the first floor under the control of DOFC2 with h = 0.02
0.15
0.1
SecondFloor Response (m)
0.05
−0.05
−0.1
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.5 Response of the second floor under the control of DOFC2 with h = 0.02
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.6 Response of the third floor under the control of DOFC2 with h = 0.02

AK BK
CK DK of controllers DOFC1 and DOFC2 as 5.5340×108 and 7.5415×104 ,
respectively. One can see clearly that the Euclidean norm of the gain of the controller
DOFC1 is about 7338 times the Euclidean norm of the gain of the controller
DOFC2, which means that a much larger control force is needed by the controller
DOFC1. Thus, we can conclude that, for the offshore steel platform, in order to
reduce the oscillation amplitudes of peak to peak of three floors to a certain range,
the introduction of a small time-delay into the output measurement can significantly
decrease the demanded control force, which is demonstrated by Figs. 7.7 and 7.8.
5
x 10
5
3
Control Forces
−1
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.7 The required control force of DOFC1
4
x 10
12
10
6
Control Forces
−2
−4
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.8 The required control force of DOFC2 with h = 0.02
Secondly, we compare the oscillation amplitudes of peak to peak of three

floors under controllers DOFC1 and DOFC2, which are listed in Table 7.1. From
this table, one can see clearly that under control of the controller DOFC1, the
oscillation amplitudes of peak to peak of the first floor, the second floor, and
the third floor are effectively reduced from 1.3793 m, 1.4946 m, and 1.5471 m
to 0.2034 m to 0.2329 m, 0.2543 m, and 0.2705 m, respectively. When a small
Table 7.1 The oscillation amplitudes of peak to peak of the three floors of the offshore platform
under no control, the control of DOFC1, and the control of DOFC2
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m)
No control 1.3793 1.4946 1.5471
DOFC1 0.2329 0.2543 0.2705
DOFC2 0.2034 0.2232 0.2391
Table 7.2 The oscillation h (s) Floor 1 (m) Floor 2 (m) Floor 3 (m)
amplitudes of peak to peak of
the three floors of the offshore 0.001 0.2037 0.2238 0.2400
platform under the control of 0.02 0.2034 0.2232 0.2391
DOFC2 for different values 0.04 0.2035 0.2229 0.2382
of time-delay h 0.08 0.2055 0.2238 0.2371
0.10 0.2076 0.2245 0.2394
0.10035 0.2365 0.2259 0.2674
0.11 1.2841 0.5904 1.5477
0.12 1.9355 0.8169 2.3421
0.2 5.1043 2.3301 5.7420
time-delay intentionally introduced in the output measurement, the designed con-

troller DOFC2 can further reduce the oscillation amplitudes of peak to peak of the
first floor, the second floor, and the third floor to 0.2034, 0.2232, and 0.2391 m,
respectively. The oscillation amplitudes of peak to peak of the first floor, the second
floor, and the third floor under the controller DOFC1 is improved by 12.67%,
12.22%, and 11.58%, respectively, under the controller DOFC2.
Thirdly, we now analyze the effects of the introduced small time-delays on the
oscillation amplitudes of peak to peak of three floors under the controller DOFC2.
Table 7.2 lists the oscillation amplitudes of peak to peak of the first floor, the second
floor, and the third floor for different values of time-delay h. From this table, it is
clear that:
(a) The oscillation amplitudes of peak to peak of the three floors can be effectively
reduced to less than 0.2438 m for h ≤ 0.1. When taking h = 0.10035, the
oscillation amplitudes of peak to peak of the three floors are the similar to the
ones derived by the controller DOFC1. However, the Euclidean norm of the
gain of the controller DOFC1 is about 13628 times the Euclidean norm of the
gain of the controller DOFC2, which means that the controller DOFC1 requires
a much larger control force than the controller DOFC2 such that the offshore
platform works in a safe environment.
(b) When h is increased to 0.11, the controlled offshore platform vibrates at a
relatively large oscillation amplitudes of peak to peak, which means that this
controller is no longer suitable for this platform. Figures 7.9, 7.10 and 7.11
depict the responses of three floors of the controlled platform when h = 0.11.
Moreover, when we continue to increase h from 0.11 to 0.2, the oscillation
amplitudes of peak to peak of the three floors are larger than the ones without
0.8
0.6
0.4
0.2
−0.2
−0.4
−0.6
−0.8
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.9 Response of the first floor under the control of DOFC2 with h = 0.11
0.3
0.2
SecondFloor Response (m)
0.1
−0.1
−0.2
−0.3
−0.4
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.10 Response of the second floor under the control of DOFC2 with h = 0.11
control. In this case, the controller will make the offshore platform work in an
unsafe environment, which is not expected.
In summary, it is shown that, when intentionally introducing a proper small
time-delay, one can design a dynamic output feedback controller such that (i) the
controller is of a small gain in the sense of Euclidean norm and (ii) the internal
oscillations of the offshore platform can be dramatically reduced.
0.8
0.6
0.4
0.2
−0.2
−0.4
−0.6
−0.8
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.11 Response of the third floor under the control of DOFC2 with h = 0.11
7.4 Conclusions
This chapter has investigated the effects of a time-delay on dynamic output feedback
control of an offshore platform subject to a nonlinear wave-induced self-excited
hydrodynamic force. It is found that by intentionally introducing a time-delay into
the output channel, the controller exhibits a small gain in the sense of Euclidean
norm, which in turn demands a correspondingly small control force. The internal
oscillation of the offshore platform is thereby dramatically reduced.
7.5 Notes
This chapter focuses on designing a delayed feedback controller for the offshore
platform by only using the measurable output signals, and the main results, to
name a few, are obtained by Zhang and Han [100]. In Nourisola et al. [103], an
adaptive output feedback integral sliding mode controller and a delayed adaptive
output feedback integral sliding mode controller have been designed for the offshore
platform. It should be mentioned that in [100] and [103], the delayed dynamic output
feedback controller and delayed adaptive output feedback sliding mode controller
are compared to the delay-free dynamic output feedback controller and delay-free
adaptive output feedback sliding mode controller, respectively. The positive effects
of artificially introduced time-delays on the active control for the offshore platform
are analyzed. Simulation results show that purposefully introduced proper time-
delays can reduce the vibration amplitudes of the system and the control cost as
well.
7.5 Notes 129
Note that the above output feedback controllers in [100, 103] and the state
feedback controllers provided by previous chapters are mainly designed in the
continuous-time domain. When a continuous-time controller for an offshore plat-
form is implemented in practice, control signals are usually transmitted in a digital
form, which results in a sampled-data system. More recently, Several sampled-data
control schemes are introduced to the area of active control for offshore platforms.
Comprehensive coverage on this subject can be found in Sakthivel et al. [109, 110],
Sivaranjani et al. [111], Zhang et al. [112], and Huang, Cai, and Xiang [113]. It is
observed that the sampled-data control schemes can reduce wave-induced vibration
of the platform significantly. Moreover, compared to continuous-time controllers,
sampled-data controllers may take less control cost. In fact, the above sampled-
data control schemes are all based on the input delay approach. In this sense,
such control schemes can also be classified as a delayed feedback control strategy,
and consequently, sampled-data control is a feasible and effective alternative
for the active control for the offshore platform. Specifically, to explore output
feedback or observer-based sampled-data control scheme for offshore platforms is
of significance both in theory and in real implementations, which is well worth
investigation in the near future.
Chapter 8
Network-Based Modeling and Active
Control
The network-based modeling and active control for an offshore steel jacket platform
with an active tuned mass damper mechanism is investigated in this chapter. A
network-based state feedback control scheme is developed. Under this scheme, the
corresponding closed-loop system is modeled by a system with an artificial interval
time-varying delay. Then, a delay-dependent stability criterion for the corresponding
closed-loop system is derived. Based on this stability criterion, a sufficient condition
on the existence of the network-based controller is obtained. It is found through
simulation results that (i) both the oscillation amplitudes of the offshore platform
and the required control force under the network-based state feedback controller
are smaller than those under the nonlinear controller [65] and the dynamic output
feedback controller [100]; (ii) the oscillation amplitudes of the offshore steel jacket
platform under the network-based feedback controller are almost the same as the
ones under the integral sliding mode controller [99], while the required control force
by the former is smaller than the one by the latter.
The rest of this chapter is organized as follows. A network-based dynamic
model of the offshore steel jacket platform is developed in Sect. 8.1. Section 8.2
is devoted to design a network-based state feedback control scheme and present
a stability criterion for the network-based offshore platform. Section 8.3 provides
some simulation results to illustrate the effectiveness of the developed control
strategies. Finally, Sect. 8.4 presents the conclusion of this chapter, and Sect. 8.5
provides a brief note.
An offshore steel jacket platform model with an active TMD mechanism shown in
Fig. 2.2 [65, 72] is considered for controller design. The dynamic equation is given
by (2.45). Let g(x, t) := Df (x, t). Then, the system (2.45) can be expressed as

https://doi.org/10.1007/978-981-13-2986-9_8
132 8 Network-Based Modeling and Active Control
ẋ(t) = Ax(t) + Bu(t) + g(x, t), t ≥ 0, x(0) = x0 (8.1)
It is assumed that the nonlinear self-excited hydrodynamic force g(x, t) on the

offshore platform is uniformly bounded and satisfies the following constraint as [65]
g(x, t) ≤ μ x (8.2)
where μ is a positive scalar.

The purpose of this chapter is to design a network-based state feedback controller
to reduce the oscillation amplitudes of the offshore platform. For this, a network-
based dynamic model of the offshore platform should be established first. In fact,
by inserting a communication network over the offshore steel jacket platform shown
in Fig. 8.1, a network-based dynamic model of the system can be obtained. In this
network-based model, an actuator is placed on the top of the offshore platform and
connected to the TMD to regulate the motion of the damper; several sensors are
placed on the TMD, the first and second floors of the offshore platform to measure
their displacements and velocities, respectively; and a network-based controller is
located in a land control station. To design the network-based active controller for
the offshore platform, the following assumptions are necessary.
Fig. 8.1 A network-based offshore structure with a TMD [120]

8.1 Problem Formulation 133
Assumption 1 The sensor is clock-driven, the controller and actuator are event-
driven.
Assumption 2 The full state variables are available for measurement, and the state
measurements are transmitted with a single packet, and no packet dropout occurs in
the transmission.
Assumption 3 The state signal x(t) is sampled as x(lh) every other h seconds,
the signal x(lh) and its time stamp l are encapsulated as a sampled-data packet
(l, x(lh)) for transmission, l = 1, 2, · · · , and h > 0 is the sampling period.
To design a network-based state feedback control law u(t) as
u(t) = Kx(kh), t ∈ [kh + τk , (k + 1)h + τk+1 ), k ∈ N (8.3)
where K is a 1×6 gain matrix to be determined, h > 0 denotes the sampling period,
and τk = τksc + τkca is the network-induced delay, which denotes the time from the
instant kh when sensor nodes sample sensor data from a plant to the instant when
actuator transfer data to the plant.
For simplicity, denote
τm = mink {τk | k ∈ N} , τM = maxk {τk | k ∈ N} (8.4)
Notice that
x(kh) = x(t − (t − kh)) (8.5)
Then, introduce an artificial time-delay d(t) in the form as
d(t) = t − kh, t ∈ [kh + τk , (k + 1)h + τk+1 ), k ∈ N (8.6)
which is a piecewise-linear continuous function satisfying

ḋ(t) = 1, t = kh + τk
(8.7)
0 ≤ τ1 ≤ d(t) ≤ τ2 < ∞, t ∈ [kh + τk , (k + 1)h + τk+1 ), k ∈ N
where
τ1 = τm , τ2 = τM + h (8.8)
Then, from (8.3), (8.4), (8.5), and (8.6), one yields the closed-loop system

ẋ(t) = Ax(t) + BKx(t − d(t)) + g(x, t), x(0) = x0
(8.9)
t ∈ [kh + τk , (k + 1)h + τk+1 ), k ∈ N
The initial condition of the state x(t) is supplemented as
x(s) = υ(s), s ∈ [−τ2 , 0] (8.10)
where υ is a continuous function with υ(0) = x0 .

Remark 8.1 It should be pointed out that by introducing an artificial delay d(t), the
corresponding closed-loop system (8.9) of the offshore platform is modeled as a
system with an interval time-varying delay.
8.2 Stability Analysis and Network-Based Controller Design
In this section, a stability criterion is derived for the network-based offshore steel
jacket platform system (8.9). Then, based on the stability criterion, a sufficient
condition on the existence of the state feedback controller (8.3) is obtained.
For simplicity, denote
η(t) := [x T (t) x T (t − d(t)) x T (t − τ1 ) x T (t − τ2 ) x T (t − ρ(t)) g T (x, t)]T

(8.11)
where I and 0 are 6 × 6 identity matrix and zero matrix, respectively, and
ρ(t) = d(t) − τk , t ∈ [kh + τk , (k + 1)h + τk+1 ), k ∈ N (8.12)
Further, denote
⎧
⎪
⎪ E1 := [I 0 0 0 0 0], E2 := [0 I 0 0 0 0]
⎨
E3 := [0 0 I 0 0 0], E4 := [0 0 0 I 0 0]
(8.13)
⎪ E5 := [0 0 0 0
⎪ I 0], E6 := [0 0 0 0 0 I ]
⎩
Eij := Ei − Ej , i, j = 1, · · · , 6
Notice
⎧
⎨ x(t) = E1 η(t), x(t − d(t)) = E2 η(t)
x(t − τ1 ) = E3 η(t), x(t − τ2 ) = E4 η(t) (8.14)
⎩
x(t − ρ(t)) = E5 η(t), g(x, t) = E6 η(t)
Then, based on (8.11) and (8.14), the closed-loop system (8.9) can be written as
ẋ(t) = Π η(t) (8.15)
where Π = AE1 + BKE2 + E6 .

As one of the main results of the paper, the following proposition presents a
stability criterion for the closed-loop system (8.9).
8.2 Stability Analysis and Network-Based Controller Design 135
Proposition 8.1 Given scalars τ1 and τ2 with τ2 > τ1 ≥ 0, μ > 0, the closed-loop
system (8.9) is asymptotically stable for d(t) satisfying (8.7) if there exist 6 × 6 real
matrices P > 0, Q1 > 0, Q2 > 0, R1 > 0, R2 > 0, S1 > 0, S2 > 0, Z, 6 × 36
matrix X, and 1 × 6 matrix K such that
⎡ ⎤
Λ11 + δΨ11 Λ12 + δΨ12 μE1T
⎣ ∗ −τ12 R1 − δ 2 R2 − δQ1 0 ⎦ < 0 (8.16)
∗ ∗ −I
⎡ ⎤
Λ11 Λ12 δXT μE1T
⎢ ∗ −τ 2 R1 − δ 2 R2 0 0 ⎥
⎢ 1 ⎥<0 (8.17)
⎣ ∗ ∗ −δQ1 0 ⎦
∗ ∗ ∗ −I

R2 Z
≥0 (8.18)
∗ R2
where δ = τ2 − τ1 and
Λ11 = Π T P E1 + E1T P Π + E13

T
S1 E13 + E34
T
S2 E34
+ E1T S1 E3 + E3T S1 E1 + E3T S2 E4 + E4T S2 E3
− E13
T
R1 E13 − E6T E6 − E32
T
R2 E32 − E24
T
R2 E24
− E32
T
ZE24 − E24
T T
Z E32 + E51
T
X + XT E51 − E15
T
Q2 E15 (8.19)
Λ12 = Π T (τ12 R1 + δ 2 R2 ) (8.20)
Ψ11 = Π T Q2 E15 + E15
T
Q2 Π, Ψ12 = Π T Q1 (8.21)
Proof Construct a Lyapunov-Krasovskii functional candidate as

4
V (xt , ẋt ) = Vj (xt , ẋt ) (8.22)
j =1
where
V1 (xt , ẋt ) =x T (t)P x(t)

t t−τ1
V2 (xt , ẋt ) = x T (s)S1 x(s)ds + x T (s)S2 x(s)ds
t−τ1 t−τ2
0 t
V3 (xt , ẋt ) =τ1 ẋ T (s)R1 ẋ(s)dsdθ
−τ1 t+θ
−τ1 t
+δ ẋ T (s)R2 ẋ(s)dsdθ
−τ2 t+θ
t
V4 (xt , ẋt ) =(τ2 − d(t)) ẋ (s)Q1 ẋ(s)ds + ψ (t)Q2 ψ(t)
T T
t−ρ(t)
with xt = x(t + θ ), θ ∈ [−τ2 , 0], P > 0, Si > 0, Ri > 0 and Qi > 0 (i = 1, 2),
and ψ(t) = x(t) − x(t − ρ(t)).
Note that for real matrices P > 0, Si > 0, Ri > 0, and Qi > 0 (i = 1, 2), on
the one hand, there exist !1 > 0 and !2 > 0 such that
!1 x(t)2 ≤ V (xt , ẋt ) ≤ !2 xt 2W (8.23)
where W is the space of functions xt (s) and ẋt (s) with the norm xt W as
xt W = sup {xt (s), ẋt (s)} (8.24)

s∈[−τ2 , 0]
On the other hand, we have
V (xkh+τk , ẋkh+τk ) ≤ lim V (xt , ẋt ), ∀k ∈ N (8.25)

t→(kh+τk )−
In fact, due to the fact that in Lyapunov-Krasovskii functional (8.22), the terms
V1 , V2 , and V3 are absolutely continuous on [0, ∞]; in addition, V4 is absolutely
continuous for t = kh + τk and does not increase along {kh + τk , k ∈ N}
since it is nonnegative before {kh + τk , k ∈ N} and becomes zero just after these
points.
In what follows, to prove the asymptotic stability of system (8.9), we only need
to prove that V̇ (xt , ẋt ) < 0 holds for any x(t) = 0 with t = kh + τk , k ∈ N. For
this, taking the derivative of the Lyapunov-Krasovskii functional (8.22) along the
trajectory of system (8.9) and noticing the fact in (8.7) yields

4
V̇ (xt , ẋt ) = V̇j (xt , ẋt ) (8.26)
j =1
where
V̇1 (xt , ẋt ) = ηT (t)[Π T P E1 + E1T P Π ]η(t) (8.27)

V̇2 (xt , ẋt ) = η T
(t)[E1T S1 E1 − E3T S1 E3 + E3T S2 E3 − E4T S2 E4 ]η(t) (8.28)
V̇3 (xt , ẋt ) = η (t)[Π
T T
(τ12 R1 + δ R2 )Π ]η(t) + χ1 (t) + χ2 (t)
2
(8.29)
V̇4 (xt , ẋt ) = ηT (t)[(τ2 − d(t))(Π T Q1 Π + Π T Q2 E15
+ E15
T
Q2 Π ) − E15
T
Q2 E15 ]η(t) + χ3 (t) (8.30)
with
t
χ1 (t) = − τ1 ẋ T (s)R1 ẋ(s)ds (8.31)
t−τ1
t−τ1
χ2 (t) = − δ ẋ T (s)R2 ẋ(s)ds (8.32)
t−τ2
t
χ3 (t) = − ẋ T (s)Q1 ẋ(s)ds (8.33)
t−ρ(t)
Note that
T
E13 S1 E13 + E34
T
S2 E34 + E1T S1 E3 + E3T S1 E1 + E3T S2 E4
+ E4T S2 E3 − (E1T S1 E1 − E3T S1 E3 + E3T S2 E3 − E4T S2 E4 )
= 2E3T S1 E3 + 2E4T S2 E4 ≥ 0 (8.34)
Then, from (8.28) and (8.34), one yields
V̇2 (xt , ẋt ) ≤ ηT (t)(E13

T
S1 E13 + E34
T
S2 E34 + E1T S1 E3
+ E3T S1 E1 + E3T S2 E4 + E4T S2 E3 )η(t) (8.35)
Applying Lemma 2.6 yields
χ1 (t) ≤ −ηT (t)E13

T
R1 E13 η(t) (8.36)
By Lemma 2.8, for any 6 × 6 matrix Z satisfying (8.18), we have
χ2 (t) ≤ −ηT (t)[E32

T
R2 E32 + E24
T
R2 E24 + E32
T
ZE24 + E24
T T
Z E32 ]η(t)
(8.37)
Similarly, by applying Lemma 2.9, the following inequality is true for any 6 × 36
matrix X:
χ3 (t) ≤ ηT (t)[E51
T
X + XT E51 + ρ(t)XT Q−1
1 X]η(t) (8.38)
Further, from (8.8) and (8.12), one gets
χ3 (t) ≤ ηT (t)[E51
T
X + XT E51 + (d(t) − τ1 )XT Q−1
1 X]η(t) (8.39)
Notice that (8.2) is equivalent to
ηT (t)(μ2 E1T E1 − E6T E6 )η(t) ≥ 0 (8.40)

Consequently, from (8.26) to (8.30), (8.35), (8.36), (8.37), (8.39), and (8.40), one
yields
V̇ (xt , ẋt ) ≤ ηT (t)[Λ11 + Λ12 (τ12 R1 + δ 2 R2 )−1 ΛT12 + μ2 E1T E1 + Δ(t)]η(t)

(8.41)
where Λ11 and Λ12 are defined in (8.19) and (8.20), respectively, and
Δ(t) = (τ2 − d(t))(Ψ11 + Ψ12 Q−1 T −1

1 Ψ12 ) + (d(t) − τ1 )X Q1 X
T
(8.42)
with Ψ11 and Ψ12 being defined by (8.21).

It is clear that if the following inequality holds
Λ11 + Λ12 (τ12 R1 + δ 2 R2 )−1 ΛT12 + μ2 E1T E1 + Δ(t) < 0 (8.43)
then there exists a scalar ! > 0 such that
V̇ (xt , ẋt ) ≤ −!ηT (t)η(t) ≤ −!x T (t)x(t) < 0, ∀x(t) = 0
which means that the system (8.9) is asymptotically stable.

Notice that Δ(t) is a convex combination of Ψ11 + Ψ12 Q−1 T T −1
1 Ψ12 and X Q1 X
on d(t) ∈ [τ1 , τ2 ]. Therefore, (8.43) holds if the following inequalities are true
Λ11 + Λ12 (τ12 R1 + δ 2 R2 )−1 ΛT12 + μ2 E1T E1 + δ(Ψ11 + Ψ12 Q−1 T

1 Ψ12 ) < 0
(8.44)
Λ11 + Λ12 (τ12 R1 + δ 2 R2 )−1 ΛT12 + μ2 E1T E1 + δXT Q−1

1 X <0 (8.45)
Due to the fact the inequality (8.44) is equivalent to
Λ11 + δΨ11 + μ2 E1T E1

+ (Λ12 + δΨ12 )(τ12 R1 + δ 2 R2 + δQ1 )−1 (Λ12 + δΨ12 )T < 0 (8.46)
Applying the Schur Complement to (8.46) and (8.45), we arrive at (8.16) and (8.17),
respectively. The proof is completed.

Based on Proposition 8.1, a sufficient condition on the existence of the network-
based state feedback controller for the system (8.9) is proposed and stated as the
following proposition.
Proposition 8.2 Given scalars τ1 and τ2 with τ2 > τ1 ≥ 0, ν1 > 0, ν2 > 0, μ > 0,
and ϑ > 0, the offshore platform system (8.1) under the networked state feedback
controller (8.3) is asymptotically stable, if there exist 6 × 6 real matrices P̄ > 0,
Q̄1 > 0, Q̄2 > 0, R̄1 > 0, R̄2 > 0, S̄1 > 0, S̄2 > 0, Z̄, 6 × 36 matrix X̄, and 1 × 6
matrix K̄ such that
⎡ ⎤
Λ̄11 + δ Ψ̄11 Λ̄12 μE1T P̄
⎣ ∗ −Ψ̄22 0 ⎦<0 (8.47)
∗ ∗ −I
⎡ ⎤
Λ̄11 Λ̄12 δ X̄T μE1T P̄
⎢ ∗ −Λ̄22 0 0 ⎥
⎢ ⎥<0 (8.48)
⎣ ∗ ∗ −δ Q̄1 0 ⎦
∗ ∗ ∗ −I

R̄2 Z̄
≥0 (8.49)
∗ R̄2
where
Λ̄11 = E1T (AP̄ E1 + B K̄E2 + P̄ E6 ) + E13

T
S̄1 E13 − E6T E6
+ (E1T P̄ AT + E2T K̄ T B T + E6T P̄ )E1 + E34
T
S̄2 E34
+ E1T S̄1 E3 + E3T S̄1 E1 + E3T S̄2 E4 + E4T S̄2 E3
− E13
T
R̄1 E13 − E32
T
R̄2 E32 − E24
T
R̄2 E24 − E32
T
Z̄E24
− E24
T T
Z̄ E32 − ϑE15
T
P̄ E15 + E51
T
X̄ + X̄T E51 (8.50)
Λ̄12 = E1T P̄ AT + E2T K̄ T B T + E6T P̄ (8.51)
Λ̄22 = 2ν1 P̄ − ν12 (τ12 R̄1 + δ 2 R̄2 ) (8.52)
Ψ̄11 = ϑ(E1T P̄ AT + E2T K̄ T B T + E6T P̄ )E15
+ ϑE15
T
(AP̄ E1 + B K̄E2 + P̄ E6 ) (8.53)
Ψ̄22 = 2ν2 P̄ − ν22 (τ12 R̄1 + δ 2 R̄2 + δ Q̄1 ) (8.54)
Moreover, the controller gain is determined by K = K̄ P̄ −1 .

Proof Denote
:= diag{P −1 , P −1 , P −1 , P −1 , P −1 , I }
Υ1 := diag{, (τ12 R1 + δ 2 R2 + δQ1 )−1 , I }
Υ2 := diag{, (τ12 R1 + δ 2 R2 )−1 , P −1 , I }
Υ3 := diag{P −1 , P −1 }
Then, let
P̄ = P −1 , K̄ = KP −1 , Q̄1 = P −1 Q1 P −1 , Q̄2 = ϑP −1
−1
X̄ = P X, −1
Ȳ = Q X, Z̄ = P −1 ZP −1
−1 −1 −1 −1
S̄i = P Si P , R̄i = P Ri P , i = 1, 2
Then pre- and post-multiplying both sides of (8.16) by Υ1 , both sides of

(8.17) by Υ2 , and both sides of (8.18) by Υ3 , respectively, applying the
inequalities
−P̄ (τ12 R̄1 + δ 2 R̄2 )−1 P̄ ≤ −2ν1 P̄ + ν12 (τ12 R̄1 + δ 2 R̄2 )
and
−P̄ (τ12 R̄1 + δ 2 R̄2 + δ Q̄1 )−1 P̄ ≤ −2ν2 P̄ + ν22 (τ12 R̄1 + δ 2 R̄2 + δ Q̄1 )
we arrive at (8.47), (8.48), and (8.49), respectively. This completes the proof.

In this section, the effectiveness of the network-based state feedback control

scheme for an offshore platform will be demonstrated first. Then, the proposed
scheme will be compared with those without network settings, such as nonlinear
control [65], dynamic output feedback control [100], and integral sliding
mode control [99]. Finally, the effects of the network-induced delays on
the network-based state feedback control for the offshore platform will be
investigated.
The data of the wave and the offshore platform with a TMD is taken from [65]
and listed in Table 4.1. From these settings, the matrices A and B in (2.45) can
be obtained as (4.34), and the nonlinear self-excited wave force g(x, t) can be
computed as appendix A in [65].
8.3.1 Performance of System with Network-Based Controllers
In this subsection, both the constant network-induced delay and the time-varying
network-induced delay are considered, and several network-based state feedback
controllers (NSFCs) will be designed, and the performance of the offshore platform
under the NSFCs will be presented.
For comparison, we first give the performance of the offshore platform without
control. As stated in [99], when no controller is utilized to control the offshore
platform, the oscillation amplitudes of the first, second, and third floors peak to peak
are 1.3738, 1.4489, and 1.5634 m, respectively. The average oscillation amplitude
of the three floors is about 1.4620 m. In this situation, the offshore platform works
in a dangerous environment.
8.3.1.1 The Constant Network-Induced Delay
Set μ = 0.8, ϑ = 0.1, and ν1 = ν2 = 10, and let the sampling period h = 0.01 s.
Suppose that the network-induced delay is a constant, i.e., τk ≡ τ, k ∈ N. In this
situation, by Proposition 8.2, set τ = 0.01, 0.02, 0.03, 0.04, and 0.05 s, respectively,
we can design different NSFCs with the form (8.3), respectively. For convenience,
the obtained controllers are denoted as NSFC1, NSFC2, NSFC3, NSFC4, and
NSFC5, respectively, and the corresponding gain matrices are listed in Table 8.1.
In what follows, unless otherwise specified, all related values of oscillation
amplitudes of the offshore platform and the control force are given in a sense of peak
to peak. Applying the above NSFCs to the offshore platform system, respectively,
the oscillation amplitudes of the first, second, and third floors of the platform and the
ranges of the control force are listed in Table 8.2, where the reduction percentages of
the average oscillation amplitudes of the offshore platform with NSFCs and without
control are also presented.
From this table, it is observed that NSFC1 can reduce the average oscillation
amplitudes of the three floors to about 13% of that of the system without control
[99]. Compared with the oscillation amplitudes of the system under other con-
trollers: NSFC2 to NSFC5, the average oscillation amplitudes under NSFC1 is
the smallest. In this case, the oscillation amplitudes of the three floors are 0.1784,
0.1969, and 0.2120 m, respectively, which can be seen from Figs. 8.2, 8.3, and 8.4.
However, it is not difficult to see from Table 8.2 and Fig. 8.5 that the control force
required by NSFC1 is about 7.6699 × 104 N, which is the largest. It indicates that
if the control cost is specifically taken into account, one may choose NSFC5 owing
to its smallest control force required, though the average vibration amplitudes of the
Table 8.1 The values of the constant network-induced delay τ , corresponding NSFCs and the
gain matrices
τ (s) Controller Gain matrix K
0.01 NSFC1 104 × [−0.2964 − 0.1050 2.7860 0.3923 0.0288 − 0.2613]
0.02 NSFC2 104 × [−0.2230 − 0.0819 1.6986 0.3564 0.0555 − 0.1996]
0.03 NSFC3 103 × [−1.6350 − 0.6521 9.1914 3.3411 0.7785 − 1.4671]
0.04 NSFC4 103 × [−0.9972 − 0.4329 0.6516 3.0963 0.9469 − 0.9254]
0.05 NSFC5 103 × [−1.0763 − 0.2700 1.4937 1.6761 0.4357 − 1.0926]
Table 8.2 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under different NSFCs via constant network-induced delays
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
No control[99] 1.3738 1.4489 1.5634 – –
NSFC1 0.1784 0.1969 0.2120 13.3 7.6699
NSFC2 0.1862 0.2053 0.2209 13.4 7.3318
NSFC3 0.1928 0.2124 0.2284 14.5 6.9681
NSFC4 0.2176 0.2391 0.2562 16.3 6.4260
NSFC5 0.2504 0.2746 0.2933 18.7 6.0955
0.2
0.15
0.1
First Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.2 The response of the first floor under NSFC1 with τk ≡ 0.01 s
0.2
0.15
0.1
Second Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.3 The response of the second floor under NSFC1 with τk ≡ 0.01 s
platform is the largest. In fact, in this case, the oscillation amplitudes of the three
floors are increased from 0.1784, 0.1969, and 0.2120 m under NSFC1 to 0.2504,
0.2746, and 0.2933 m, respectively. The average oscillation amplitudes of the three
floors under NSFC5 is about 18.7% of that of the offshore platform without control.
Therefore, choosing a network-based controller among NSFC1 to NSFC5, there
exists a tradeoff between control performance and cost.
0.2
0.15
0.1
Third Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.4 The response of the third floor under NSFC1 with τk ≡ 0.01 s
4
x 10
10
6
Control Force (N)
−2
−4
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.5 The response of the control force under NSFC1 with τk ≡ 0.01 s
8.3.1.2 The Time-Varying Network-Induced Delay
It is assumed that the time-varying network-induced delay τk varies on the interval

[τm , τM ]. For some given different values of τm and τM , we tend to design several
NSFCs to investigate the effectiveness of the proposed network-based state feedback
control scheme.
Let τm and τM be taken values as the following cases:
Case I: τm = 0.01 s, τM = 0.02 s

Case II: τm = 0.02 s, τM = 0.03 s
Case III: τm = 0.03 s, τM = 0.04 s
Case IV: τm = 0.01 s, τM = 0.03 s
Case V: τm = 0.02 s, τM = 0.04 s
Case VI: τm = 0.01 s, τM = 0.04 s
The values of μ, ϑ, ν1 , ν2 , and h are set as the ones as Sect. 8.3.1.1. Then, in
each case, by Proposition 8.2, we can design corresponding NSFC, respectively. The
obtained network-based controllers are denoted by NSFC6 to NSFC11, respectively,
and the values of gain matrix K are presented by Table 8.3. When the obtained
controllers NSFC6 to NSFC11 are utilized to control the offshore platform, the
oscillation amplitudes of the first, second, and third floors, the reduction percentages
of the average oscillation amplitudes of the three floors with NSFCs and without
control, and the ranges of the control force are given in Table 8.4.
Table 8.3 The bounds τm and τM of the time-varying network-induced delay τk , corresponding
NSFCs and the gain matrices
τm (s) τM (s) Controller Gain matrix K
τM − τm = 0.01 (s)
0.01 0.02 NSFC6 104 × [−0.2324 −0.0864 1.8577 0.3647 0.0531 −0.2065]
0.02 0.03 NSFC7 103 × [−1.6052 −0.6461 8.8364 3.3530 0.7898 −1.4439]
0.03 0.04 NSFC8 103 × [−0.9980 −0.4346 0.6806 3.1036 0.9465 −0.9292]
τM − τm = 0.02 (s)
0.01 0.03 NSFC9 103 × [−1.6343 −0.6547 9.2164 3.3560 0.7805 −1.4688]
0.02 0.04 NSFC10 103 × [−0.9912 −0.4318 0.5843 3.1071 0.9479 −0.9228]
τM − τm = 0.03 (s)
0.01 0.04 NSFC11 103 × [−0.9925 −0.4321 0.6120 3.0988 0.9469 −0.9239]
ranges of control force under different NSFCs via time-varying network-induced delays
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
No control[99] 1.3738 1.4489 1.5634 – –
NSFC6 0.1833 0.2022 0.2176 13.8 7.4175
NSFC7 0.1933 0.2130 0.2290 14.5 6.9496
NSFC8 0.2177 0.2392 0.2564 16.2 6.4309
NSFC9 0.1928 0.2124 0.2284 14.5 6.9737
NSFC10 0.2181 0.2396 0.2568 16.3 6.4306
NSFC11 0.2181 0.2396 0.2567 16.3 6.4394
From this table, one can see that the designed NSFCs can be classified two
groups. The first one includes NSFC8, NSFC10, and NSFC11, under which the
average oscillation amplitudes of the three floors can be reduced to about 16% of
the ones without control, and the average range of the control force by NSFC8,
NSFC10, and NSFC11 is less than 6.5 ×104 N. The second one involves NSFC6,
NSFC7, and NSFC9, and under these three controllers, the average oscillation
amplitudes of the system can be reduced to even about 14% of those when no
controller is used, while the average range of the control force required is 7.2
×104 N, which is larger than the one by the NSFCs in the first group. Specifically,
as representative controllers of the two groups, under NSFC 7 and NSFC11, the
response curves of the three floors and the control force are demonstrated by
Figs. 8.6, 8.7, 8.8, 8.9, 8.10, 8.11, 8.12, and 8.13, respectively.
8.3.2 Comparison of Controllers With and Without Network

Setting
Note the fact from Tables 8.2 and 8.4 that under different NSFCs, if the controlled
oscillation amplitudes of the three floors of the offshore platform are smaller,
the control force required is generally larger. For instance, under NSFC1, the
oscillation amplitudes of the floors are the smallest, while the control force is
the largest, and under NSFC5, the required control cost is the smallest, while
the control performance is not the best. Therefore, to compare the network-based
control scheme with some existing control schemes without network setting under
0.2
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.6 The response of the first floor under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
0.2
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.7 The response of the second floor under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
0.2
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.8 The response of the third floor under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
a relative fair condition, we select three network-based controllers: NSFC3, NSFC7

and NSFC11 under which the oscillation amplitudes of the floors are not the best,
and the ranges of the control force required are not the largest.
Table 8.5 lists the ranges of the required control force, the oscillation amplitudes
of the three floors of the offshore platform without control, under nonlinear
controller (NLC) [65], dynamic output feedback controller (DOFC) [100], sliding
mode controller (SMC) [99], NSFC3, NSFC7, and NSFC3. In this table, the
4
x 10
10
6
Control Force (N)
−2
−4
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.9 The response of the control force under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
0.2
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.10 The response of the first floor under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
reduction percentages of the average oscillation amplitudes of the three floors of

the offshore platform with controllers and without control are also listed to compare
their effectiveness. It can be observed from Table 8.5 that the following facts are
true:
0.2
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.11 The response of the second floor under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
0.2
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.12 The response of the third floor under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
• Comparison of NSFC3, NSFC7 with NLC, DOFC, and SMC The oscillation
amplitudes of the three floors of the offshore platform under NSFC3 and NSFC7
are smaller than the ones under NLC, DOFC, and SMC. Moreover, the ranges of
the control force required by NSFC3 and NSFC7 are much smaller than the ones
by NLC, DOFC and SMC. In fact, under NLC, DOFC and SMC, the average
oscillation amplitudes of the three floors are reduced to about 21%, 17% and
4
x 10
10
4
Control Force (N)
−2
−4
−6
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.13 The response of the control force under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
ranges of control force under controllers without and with network settings
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) u (104 N)
NLC [65] 0.3050 0.3050 0.3050 20.9 20.000
DOFC [100] 0.2329 0.2543 0.2705 17.3 40.000
SMC [99] 0.2192 0.2301 0.2383 15.7 21.565
NSFC3 0.1928 0.2124 0.2284 14.5 6.9681
NSFC7 0.1933 0.2130 0.2290 14.5 6.9496
NSFC11 0.2181 0.2396 0.2567 16.3 6.4394
16% of those of the platform without control, respectively, while the average
oscillation amplitudes of the floors can be reduced even to about 14% under
NSFC3 and NSFC7 of their original value. In addition, the ranges of the control
force by NLC, DOFC and SMC are about 2.9, 5.8 and 3.1 times as the ones by
NSFC3 and NSFC7.
• Comparison of NSFC11 with NLC, DOFC, and SMC On the one hand, both
the oscillation amplitudes of the three floors and required control force under
NSFC11 are much smaller than the ones under NLC and DOFC, respectively.
This shows that NSFC11 is better than NLC and DOFC from the perspective of
the control performance and cost. On the other hand, the oscillation amplitudes of
the three floors under NSFC11 are almost in the same level as the ones by SMC,
while the control force required by SMC is 3.3 times as the one by NSFC11,
which indicates that to obtain the same control performance, the control cost
required by NSFC11 is smaller the one by SMC.
8.3.3 Effect Analysis of Network-Induced Delays
This subsection focuses on investigating the effect of the network-induced delay

on the network-based state feedback control for the offshore platform. For the
network-based controllers: NSFC3, NSFC7, and NSFC11 designed in Sect. 8.3.1,
we will discuss the effect of the network-induced delay on the control of the offshore
platform from the oscillation amplitudes of the three floors and the required control
force, respectively.
First, under assumption that the network-induced delay is invariant, i.e., τk ≡
τ, k ∈ N, in this case, for NSFC3, let the network-induced delay τ varies from
0.01 to 0.06 s with a step 0.01 s. Then, under NSFC3, the oscillation amplitudes of
the three floors, the average amplitude reductions of the offshore platform, and the
range of the control force are computed and listed in Table 8.6.
Second, suppose that the network-induced delay τk , k ∈ N is time-varying,
and τk ∈ [τm , τM ]. In this case, let the minimum τm of the network-induced delay
is given, while the maximum τM of the delay increases with a step 0.01 s. Then
we investigate the oscillation amplitudes of the offshore platform and the required
control force under NSFC7 and NSFC11, respectively. Under NSFC7, let τm is taken
as 0.01, 0.02, 0.03, and 0.04 s. Then for each given value of τm , let τM varies from
τm + h to 0.07 s, the oscillation amplitudes of the three floors and the control force
are presented in Table 8.7. Similarly, Table 8.8 gives the oscillation amplitudes
of the first, second and third floors of the offshore platform and the ranges of the
control force under NSFC11, where τm is set as 0.01, 0.02, and 0.03 s, respectively,
and in each case of τm , the maximum τM of the network-induced delay gradually
increases to 0.09 s from τm + h. From Tables 8.6, 8.7, and 8.8, it can be observed
that:
• The effect of the network-induced delay on the control performance of the
offshore platform and the control cost As the network-induced delay τ tends
to large in the case of NSFC3, or for the given minimum network-induced
delay τm , if the maximum network-induced delay τM tends to become large
in the cases of NSFC7 and NSFC11, the average oscillation amplitudes of the
three floors of the offshore platform are become large slightly, while the control
ranges of control force under NSFC3 as the network-induced delay τk ≡ τ varies
τ (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
0.01 0.1925 0.2121 0.2281 14.3 6.9665
0.02 0.1927 0.2123 0.2282 14.4 6.9627
0.03 0.1928 0.2124 0.2284 14.4 6.9680
0.04 0.1929 0.2125 0.2285 14.5 6.9843
0.05 0.1930 0.2126 0.2286 14.6 7.0014
0.06 0.1931 0.2127 0.2287 14.6 6.9958
ranges of control force under NSFC7 as τk varies on interval [τm , τM ]
τm (s) τM (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
0.01 0.02 0.1932 0.2128 0.2288 14.5 6.9485
0.03 0.1933 0.2130 0.2290 14.5 6.9592
0.04 0.1934 0.2131 0.2291 14.5 6.9729
0.05 0.1935 0.2132 0.2292 14.5 6.9875
0.06 0.1935 0.2132 0.2292 14.5 7.0319
0.07 0.1936 0.2133 0.2293 14.5 7.4784
0.02 0.02 0.1932 0.2128 0.2288 14.5 6.9419
0.03 0.1933 0.2130 0.2290 14.5 6.9558
0.04 0.1934 0.2131 0.2291 14.5 6.9742
0.05 0.1935 0.2132 0.2292 14.5 6.9903
0.06 0.1936 0.2132 0.2292 14.5 7.0027
0.07 0.1938 0.2133 0.2293 14.5 13.191
0.03 0.03 0.1933 0.2130 0.2290 14.5 6.9474
0.04 0.1934 0.2131 0.2291 14.5 6.9734
0.05 0.1935 0.2132 0.2292 14.5 6.9891
0.06 0.1936 0.2133 0.2293 14.5 7.0332
0.07 0.1937 0.2134 0.2300 14.5 14.230
0.04 0.04 0.1934 0.2131 0.2291 14.5 6.9634
0.05 0.1935 0.2132 0.2292 14.5 6.9902
0.06 0.1936 0.2133 0.2293 14.5 7.0099
0.07 0.1936 0.2139 0.2339 14.5 74.391
force required tends to become large significantly. It means that the larger the
network-induced delay, the more control cost NSFC3, NSFC7, and NSFC11 will
take.
• The allowable maximum network-induced delay The allowable maximum
network-induced delay under NSFC3, NSFC7, and NSFC11 is 0.06, 0.07, and
0.08 s, respectively. It means that if the network-induced delay is larger than
0.06 s for NSFC3, 0.07 s for NSFC7, and 0.08 s for NSFC11, respectively, the
controlled oscillation amplitudes of the three floors or the required control force
may increase significantly, in these cases, the obtained controllers may not work.
Also, it should be pointed herein that the allowable maximum network-induced
delay under NSFC11 is larger than the one under NSFC3, which means that
NSFC11 is more practical than both NSFC3 and NSFC7.
In summary, based on the developed network-based dynamic model of the
offshore platform, for the proper network-induced delays, the network-based state
feedback control scheme can significantly attenuate the vibration of the offshore
platform. In addition, the proposed network-based control scheme is superior to
some existing control strategies without network-setting, i.e., nonlinear control [65],
dynamic output feedback control [100], and integral sliding mode control [99].
ranges of control force under NSFC11 as τk varies on interval [τm , τM ]
τm (s) τM (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
0.01 0.02 0.2175 0.2390 0.2561 16.3 6.3898
0.03 0.2178 0.2393 0.2564 16.3 6.4177
0.04 0.2181 0.2396 0.2567 16.3 6.4287
0.05 0.2184 0.2399 0.2570 16.3 6.4448
0.06 0.2186 0.2401 0.2572 16.3 6.4763
0.07 0.2188 0.2403 0.2574 16.3 6.5687
0.08 0.2192 0.2406 0.2578 16.4 7.3361
0.09 0.2203 0.2409 0.2606 16.5 80.256
0.02 0.03 0.2178 0.2393 0.2564 16.3 6.4154
0.04 0.2181 0.2396 0.2567 16.3 6.4272
0.05 0.2184 0.2399 0.2570 16.3 6.4448
0.06 0.2186 0.2402 0.2573 16.3 6.4707
0.07 0.2189 0.2404 0.2575 16.3 6.5943
0.08 0.2192 0.2406 0.2577 16.4 7.4523
0.09 0.2216 0.2409 0.2601 16.5 137.32
0.03 0.04 0.2181 0.2396 0.2567 16.3 6.4331
0.05 0.2184 0.2399 0.2570 16.3 6.4442
0.06 0.2187 0.2402 0.2573 16.3 6.4777
0.07 0.2189 0.2405 0.2576 16.4 6.5940
0.08 0.2193 0.2407 0.2578 16.4 7.4242
0.09 0.2315 0.2425 0.2311 17.0 415.94
8.4 Conclusions
In this chapter, a network-based dynamic model of the offshore platform has been
established and a network-based control scheme for the offshore platform has been
developed. A new Lyapunov-Krasovskii functional has been constructed to derive
a delay-dependent sufficient condition, which makes the system asymptotically
stable. Based on the condition, the gain matrix of the network-based state feedback
controller has been obtained. The simulation results have demonstrated that the
proposed network-based controller is effective to improve the control performance
of the offshore platform and reduce the required control cost. Moreover, the
proposed network-based state feedback controllers are better than some existing
ones without network setting.
8.5 Notes
Note that networked control system is connected over networked media, it is a viable
and efficient way to deal with remote control problems. In recent two decades, net-
worked control system has received increasing research interest, a comprehensive
8.5 Notes 153
coverage of this topic can be found in Ge et al. [117], Zhang et al. [119], and the
references therein. So far, there have been considerable research results concerning
the modeling, stability analysis and control, and filtering problems for the networked
system [118]. In addition, network-based control strategies have been applied to
several engineering applications.
Due to the fact that offshore structures are often far from land and always
affected by very complicated and harsh ocean environmental loads. From the
saving low control cost point of view, simplifying installation and maintenance,
improving reliability, and increasing safety of the staff on the structure, it is a
feasible and efficient way to control the offshore structure over the network media.
Based on [120], this chapter firstly provides a networked dynamic model and
then presents a network-based active controller for the offshore platform with a
TMD mechanism. The superiority of network-based controller over some existing
ones without network setting such as the nonlinear controller [65], the dynamic
output feedback controller [100], and the integral sliding mode controller [99]
is investigated. However, in [120], only network-induced delays are investigated,
while other network-induced factors including packet dropouts, packet disorders,
quantization errors, and network congestion are ignored. Therefore, by taking into
account one or more of network-induced delays, packet dropouts, packet disorders,
quantization errors, and network congestion, to establish more general network-
based dynamic models and design effective network-based controllers for offshore
platform systems is still a challenging topic.
Chapter 9
Event-Triggered H∞ Reliable Control
in Network Environments
This chapter investigates the network-based modeling and event-triggered H∞

reliable control for an offshore structure. First, a network-based model of the
offshore structure subject to external wave force and actuator faults is presented.
Second, an event-triggering mechanism is proposed such that during the control
implementation, only requisite sampled-data is transmitted over networks. Third, an
event-triggered H∞ reliable control problem for the offshore structure is solved by
employing the Lyapunov-Krasovskii functional approach, and the desired controller
can be derived. It is shown through simulation results that for possible actuator
failures, the networked controller is capable of guaranteeing the stability of the
offshore structure. In addition, compared with the H∞ control scheme without
network settings, the proposed controller can suppress the vibration of the offshore
structure to almost the same level as the H∞ controller, while the former requires
less control cost. Furthermore, under the network-based controller, the communica-
tion resources can be saved significantly.
This chapter is organized as follows. Section 9.1 formulates the purpose of the
paper. Section 9.2 provides an approach to design event-triggered H∞ reliable
controllers which ensure the performance of the closed-loop system. Section 9.3
presents the simulation results. Finally, the findings are concluded in Sect. 9.4, and
a note is given in Sect. 9.5.
An offshore structure with AMD mechanisms considered in this chapter is shown in

Fig. 2.1 [79], where the first dominant vibration mode of the structure is taken into
account. The dynamic equations of the offshore structure is presented by (2.3).

https://doi.org/10.1007/978-981-13-2986-9_9
156 9 Event-Triggered H∞ Reliable Control in Network Environments
Fig. 9.1 Control diagram for a network-based offshore structure with an AMD [123]
The control output equation is given as
z(t) = C1 x(t) + D1 f (t) (9.1)
where C1 and D1 are real matrices with appropriate dimensions.

To control the offshore structure in the network setting, a communication net-
work is introduced for the offshore structure. The network-based control diagram for
the offshore structure is presented by Fig. 9.1, where an event-triggered mechanism
is applied to save the communication resources. To obtain the main results, the
necessary assumptions are given as follows.
To design the event-triggered H∞ reliable controller for the offshore platform,
Assumptions 1, 2, and 3 of Chap. 8 are necessary.
In the event-triggering mechanism under consideration, the last released data
(ik , x(ik h)), ik = 1, 2, · · · is storied by the event-trigger and utilized to check
if the following triggering condition is satisfied:
1 1 1
Ω 2 (x(ik h + j h) − x(ik h)) ≤ σ 2 Ω 2 x(ik h) (9.2)
where (ik + j, x(ik h + j h)), j = 1, 2, · · · denotes the current packet, σ > 0 is a

threshold and Ω > 0 is a weighting matrix.
During the sampling process, only in the case that the triggering condition (9.2)
does not meet, the current packet is immediately transmitted to the zero-order-hold
(ZOH) through the network. If the ZOH receives the data packet, it immediately
sends the packet to the controller. By considering the network-induced delays, the
time instant when the packet (ik , x(ik h)) reaches the ZOH is expressed as tk , k =
1, 2, · · · . Then, the network-induced delay can be expressed as
τk := tk − ik h, k = 1, 2, · · · (9.3)
Considering the possible actuator faults in the offshore structure, we design a

network-based active controller as
u(t) = r(t)Kxh (t) = r(t)Kx(ik h), t ∈ [tk , tk+1 ) (9.4)

9.1 Problem Formulation 157
where K ∈ R1×4 is a controller gain matrix to be determined, r(t) represents an

actuator-fault-related time-varying function satisfying
0 ≤ rm ≤ r(t) ≤ rM < ∞ (9.5)
where rm and rM are known real constants.

Denote
rm + rM r(t) − r̄
r̄ = , r1 (t) = (9.6)
2 r̄
One yields
r(t) = r̄(1 + r1 (t)) (9.7)
It is clear that |r1 (t)| ≤ 1.

Remark 9.1 In (9.7), the values r̄ and r1 (t) can express different fault states of
an actuator. In fact, if r̄ = 1 and r1 (t) ≡ 0 in (9.7), there is no any fault in the
actuator. If r̄ = 0, the actuator is in a total loss-of-effectiveness condition. This
case may include a lock-in-place fault, a float fault, and even an outage. In other
cases of r̄ and r1 (t), certain faults occur in the actuator and lead to the actuator
in a partial loss-of-effectiveness state. In this situation, due to the component wear,
aging, external disturbance or other factors, the output signal of the actuator deviates
from its nominal value.
Denote
ϕk := min{j | tk + j h ≥ tk+1 }, j = 0, 1, 2, · · · (9.8)
It is clear that ϕk ≥ 1. Then, similar to [100], we denote the interval [tk , tk+1 ] as
'
ϕk
[tk , tk+1 ] = ϑj (9.9)
j =1
where
ϑj = [tk + (j − 1)h, tk + j h), j = 1, 2, · · · , ϕk − 1

ϑϕk = [tk + (ϕk − 1)h, tk+1 )
Now, we introduce an artificial time-delay ρ(t) on [tk , tk+1 ) as

⎧
⎪
⎪ t − ik h, t ∈ ϑ1
⎪
⎨ t − ik h − h, t ∈ ϑ2
ρ(t) = .. .. (9.10)
⎪
⎪ . .
⎪
⎩
t − ik h − (ϕk − 1)h, t ∈ ϑϕk
It can be obtained that the delay function ρ(t) satisfies
0 ≤ τm ≤ τk ≤ ρ(t) < h + τk ≤ h + τM (9.11)
where
τm = min{τk |k = 1, 2, · · · }, τM = max{τk |k = 1, 2, · · · } (9.12)
Similarly, we define a state error function δ(t) over [tk , tk+1 ) as follows:
⎧
⎪
⎪ x(ik h) − x(ik h), t ∈ ϑ1
⎪
⎨ x(ik h) − x(ik h + h), t ∈ ϑ2
δ(t) = .. .. (9.13)
⎪
⎪ . .
⎪
⎩
x(ik h) − x(ik h + (ϕk − 1)h), t ∈ ϑϕk
Then, the triggering condition (9.2) can be written as
δ T (t)Ωδ(t) ≤ σ [δ T (t) + x T (t − ρ(t))]Ω[δ(t) + x(t − ρ(t))] (9.14)
and the input signal xh (t) in (9.4) can be rewritten as
xh (t) = δ(t) + x(t − ρ(t)), t ∈ [tk , tk+1 ) (9.15)
Further, from (9.15), the controller (9.4) can be expressed as
u(t) = r(t)K(δ(t) + x(t − ρ(t))), t ∈ [tk , tk+1 ) (9.16)
Substituting (9.16) into (2.3), we obtain the closed-loop system as
ẋ(t) = Γ1 (t)α(t) (9.17)
where

Γ1 (t) = A r(t)BK 0 0 r(t)BK D (9.18)
T
α(t) = x(t) x(t − ρ(t)) x(t − τm ) x(t − η̄) δ(t) f (t) (9.19)
This paper aims at designing an event-triggered H∞ reliable controller (9.16) such

that (i) the network-based system (9.17) with (9.1) is asymptotically stable and (ii)
the H∞ performance
z(t) < γ f (t) (9.20)
of the closed-loop system can be ensured for the wave force f (t) ∈ L2 [0, ∞) and a
prescribed γ > 0.
9.2 Design of an Event-Triggered H∞ Reliable Controller 159
9.2 Design of an Event-Triggered H∞ Reliable Controller
In this section, a design approach to the event-triggered H∞ reliable controller will

be developed. By using the obtained stability criteria, some sufficient conditions
are derived to develop the event-triggered H∞ reliable controller and the event-
triggering communication scheme.
For sufficient conditions of the existence of the gain matrix K in (9.16), we have
following proposition.
Proposition 9.1 For given scalars τm and τM satisfying 0 ≤ τm ≤ τM , r̄ > 0, γ >
0 and h > 0, the closed-loop system (9.17) with f (t) = 0 is asymptotically stable
and the H∞ performance (9.20) is guaranteed for the wave force f (t) ∈ L2 [0, ∞]
and the prescribed γ , if there exist 4×4 matrices P > 0, Q1 > 0, Q2 > 0, R1 > 0,
R2 > 0, S, Ω > 0, a 1 × 4 matrix K, and a positive scalar ε such that

R2 S
≥0 (9.21)
∗ R2
⎡ ⎤
Δ11 Δ12 Δ13
⎣ ∗ Δ22 Δ23 ⎦ < 0 (9.22)
∗ ∗ Δ33
where
⎡ ⎤
Θ r̄P BK R1 0 r̄P BK P D
⎢ ∗ −2R + S + S T R2 − S R2 − S T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q2 − Q1 − R1 − R2 S T 0 0 ⎥
Δ11 =⎢ ⎥
⎢∗ ∗ ∗ −Q2 − R2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
(9.23)
⎡ ⎤ ⎡ ⎤
0 τm AT (η̄ − τm )AT C1T εP B 0
⎢ I r̄τ K T B T r̄(η̄ − τ )K T B T 0 ⎥ ⎢ 0 r̄K T ⎥
⎢ m m ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
Δ12 =⎢ ⎥ , Δ13 = ⎢ ⎥ (9.24)
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ I r̄τm K T B T r̄(η̄ − τm )K T B T 0 ⎦ ⎣ 0 r̄K T ⎦
0 τm D T (η̄ − τm )D T D1T 0 0
Δ22 = diag{−(σ Ω)−1 , −R1−1 , −R2−1 , −I } (9.25)

T
0 ετm B T ε(η̄ − τm )B T 0
Δ23 = (9.26)
0 0 0 0
Δ33 = diag{−εI, −εI } (9.27)
with Θ = P A + AT P + Q1 − R1 and η̄ = τM + h.
Proof Construct a new Lyapunov-Krasovskii functional candidate as
V (t, x(t)) = V1 (t, x(t)) + V2 (t, x(t)) + V3 (t, x(t)) (9.28)
where
V1 (t, x(t)) =x T (t)P x(t)

t t−τm
V2 (t, x(t)) = x T (s)Q1 x(s)ds + x T (s)Q2 x(s)ds
t−τm t−η̄
0 t
V3 (t, x(t)) =τm ẋ T (s)R1 ẋ(s)dsdθ
−τm t+θ
−τm t
+ (η̄ − τm ) ẋ T (s)R2 ẋ(s)dsdθ
−η̄ t+θ
where P > 0, Q1 > 0, Q2 > 0, R1 > 0, and R2 > 0.

Taking the derivative of V (t, x(t)) along the trajectory of (9.17) yields
V̇ (t, x(t)) = V̇1 (t, x(t)) + V̇2 (t, x(t)) + V̇3 (t, x(t)) (9.29)
where
V̇1 (t, x(t)) =x T (t)(P A + AT P )x(t)

+ 2x T (t)P [r(t)BKx(t − ρ(t)) + r(t)BKδ(t) + Df (t)] (9.30)
V̇2 (t, x(t)) =x T (t)Q1 x(t) − x T (t − η̄)Q2 x(t − η̄)
+ x T (t − τm )(Q2 − Q1 )x(t − τm ) (9.31)
t
V̇3 (t, x(t)) =ẋ T (t)[τm2 R1 + (η̄ − τm )2 R2 ]ẋ(t) − τm ẋ T (s)R1 ẋ(s)ds
t−τm
t−τm
− (η̄ − τm ) ẋ T (s)R2 ẋ(s)ds (9.32)
t−η̄
From (9.17), one yields
ẋ T (t)[τm2 R1 + (η̄ − τm )2 R2 ]ẋ(t)

= α T (t)[τm2 Γ1T (t)R1 Γ1 (t) + (η̄ − τm )2 Γ1T (t)R2 Γ1 (t)]α(t) (9.33)
Using Jensen inequality yields

t
−τm ẋ T (s)R1 ẋ(s)ds ≤ −[x T (t)−x T (t −τm )]R1 [x(t)−x(t −τm )] (9.34)
t−τm
Note that matrices R2 and S satisfy the constraint (9.21). Then, by Lemma 2.8,
we have
t−τm
− (η̄ − τm ) ẋ T (s)R2 ẋ(s)ds
t−η̄
≤ − [x (t − ρ(t)) − x T (t − η̄)]R2 [x(t − ρ(t)) − x(t − η̄)]

T
− [x T (t − τm ) − x T (t − ρ(t))]R2 [x(t − τm ) − x(t − ρ(t))]

− [x T (t − ρ(t)) − x T (t − η̄)]S[x(t − τm ) − x(t − ρ(t))]
− [x T (t − τm ) − x T (t − ρ(t))]S T [x(t − ρ(t)) − x(t − η̄)] (9.35)
On the other hand, the triggering condition (9.36) can be rewritten as
δ T (t)Ωδ(t) ≤ α T (t)Γ2T (σ Ω)Γ2 α(t) (9.36)

where Γ2 = 0 I 0 0 I 0 .
Then, from (9.29), (9.30), (9.31), (9.32), (9.33), (9.34), (9.35), and (9.36), the
following matrix inequality is true:
V̇ (t, x(t)) ≤ α T (t)[Π1 (t) + Π2 (t)]α(t) (9.37)
where
⎡ ⎤
Θ r(t)P BK R1 0 r(t)P BK PD
⎢ ∗ −2R + S + S T − − T 0 ⎥
⎢ 2 R 2 S R2 S 0 ⎥
⎢ ⎥
⎢ ∗ ∗ Q 2 − Q 1 − R 1 − R2 S T 0 0 ⎥
Π1 (t) = ⎢ ⎥
⎢∗ ∗ ∗ −Q2 − R2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω 0 ⎦
∗ ∗ ∗ ∗ ∗ 0
Π2 (t) = Γ2T (σ Ω)Γ2 + τm2 Γ1T (t)R1 Γ1 (t) + (η̄ − τm )2 Γ1T (t)R2 Γ1 (t)
Now, we turn to prove that the system (9.17) is asymptotically stable. For this,
set f (t) = 0 in (9.17) and denote

Λ1 (t) = A r(t)BK 0 0 r(t)BK
T
β(t) = x(t) x(t − ρ(t)) x(t − τm ) x(t − η̄) δ(t)
Then, one can write the system (9.17) as
ẋ(t) = Λ1 (t)β(t) (9.38)

Correspondingly, the inequality (9.37) can be simplified as
V̇ (t, x(t)) ≤ β T (t)[Ξ1 (t) + Ξ2 (t)]β(t) (9.39)
where
⎡ ⎤
Θ r(t)P BK R1 0 r(t)P BK
⎢ ∗ −2R + S + S T R2 − S R2 − S T ⎥
⎢ 2 0 ⎥
⎢ ⎥
Ξ1 (t) = ⎢ ∗ ∗ Q2 − Q1 − R1 − R2 S T 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ −Q2 − R2 0 ⎦
∗ ∗ ∗ ∗ −Ω
(9.40)
Ξ2 (t) = ΛT2 (σ Ω)Λ2 + τm2 ΛT1 (t)R1 Λ1 (t) + (η̄ − τm )2 ΛT1 (t)R2 Λ1 (t)

with Λ2 = 0 I 0 0 I .
To guarantee that the system (9.38) is asymptotically stable, we require that there
exist 4 × 4 matrices P > 0, Q1 > 0, Q2 > 0, R1 > 0, R2 > 0, S, Ω > 0, and a
1 × 4 matrix K such that
V̇ (t, x(t)) ≤ −κx T (t)x(t) < 0, ∀x(t) = 0 (9.41)
where κ > 0. In order to ensure (9.41), we require the following condition
Ξ1 (t) + Ξ2 (t) < 0 (9.42)
For simplicity, let

Φ11 Φ12
Φ= (9.43)
∗ Φ22
where
⎡ ⎤
Θ r̄P BK R1 0 r̄P BK
⎢ ∗ −2R + S + S T R2 − S R2 − S T 0 ⎥
⎢ 2 ⎥
⎢ ⎥
Φ11 =⎢∗ ∗ Q2 − Q1 − R1 − R2 ST 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ −Q2 − R2 0 ⎦
∗ ∗ ∗ ∗ −Ω
⎡ ⎤
0 τm AT (η̄ − τm )AT
⎢ I r̄τ K T B T r̄(η̄ − τ )K T B T ⎥
⎢ m m ⎥
⎢ ⎥
Φ12 = ⎢0 0 0 ⎥
⎢ ⎥
⎣0 0 0 ⎦
I r̄τm K T B T r̄(η̄ − τm )K T B T
Φ22 = diag{−(σ Ω)−1 , −R1−1 , −R2−1 }
Then, applying the Schur complement to (9.42) gets

Ξ1 (t) Ξ3 (t)
<0 (9.44)
∗ Φ22
where Ξ1 (t) is given by (9.40), and

⎡ ⎤
0 τm AT (η̄ − τm )AT
⎢ I r(t)τ K T B T r(t)(η̄ − τ )K T B T ⎥
⎢ m m ⎥
⎢ ⎥
Ξ3 (t) = ⎢ 0 0 0 ⎥
⎢ ⎥
⎣0 0 0 ⎦
I r(t)τm K T B T r(t)(η̄ − τm )K T B T
Note that the inequality (9.44) can be further written as
Φ + r1 (t)1 2 + r1 (t)T2 T1 < 0 (9.45)
where Φ is determined by (9.43), and
T
1 = B T P 0 0 0 0 0 τm B T (η̄ − τm )B T

2 = 0 r̄K 0 0 r̄K 0 0 0
By S -procedure and Schur complement, for any ε > 0, the matrix inequality
(9.45) is equivalent to
⎡ ⎤
Φ ε1 T2
⎣ ∗ −εI 0 ⎦ < 0 (9.46)
∗ ∗ −εI
Due to the fact that the inequalities (9.21) and (9.22) hold, it is clear that the above
inequality is true, which means that the system (9.17) is asymptotically stable.
Now, we focus on proving that the H∞ performance (9.20) is guaranteed for the
external wave force f (t) under
zero initial condition.
Let Γ3 = C1 0 0 0 0 D1 . Then, from (9.1) and (9.37), one yields
V̇ (t, x(t)) + zT (t)z(t) − γ 2 f T (t)f (t) ≤ α T (t)Π (t)α(t) (9.47)

where
⎡ ⎤
Θ r(t)P BK R1 0 r(t)P BK P D
⎢ ∗ −2R + S + S T R2 − S R2 − S T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q2 − Q1 − R1 − R2 S T 0 0 ⎥
Π (t) = ⎢ ⎥
⎢∗ ∗ ∗ −Q2 − R2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
+ Γ2T (σ Ω)Γ2 + τm2 Γ1T (t)R1 Γ1 (t) + (η̄ − τm )2 Γ1T (t)R2 Γ1 (t) + Γ3T Γ3
(9.48)
In what follows, we prove that Π (t) < 0 is true, if matrix inequalities (9.21) and
(9.22) hold. In fact, by Schur complement, Π (t) < 0 if and only of the following
inequality is true:

Δ11 Δ12
+ r1 (t)Σ1 Σ2 + r1 (t)Σ2T Σ1T < 0 (9.49)
∗ Δ22
where
T
Σ1 = B T P 0 0 0 0 0 0 τm B T (η̄ − τm )B T 0

Σ2 = 0 r̄K 0 0 r̄K 0 0 0 0 0
By S -procedure and the Schur complement again, the known conditions (9.21) and
(9.22) guarantee that the inequality (9.49) holds. Then from (9.47), one yields
V̇ (t, x(t)) + zT (t)z(t) − γ 2 f T (t)f (t) < 0 (9.50)
Note that V (t, x(0)) = 0 under zero initial condition. Integrating both sides of
(9.50) from 0 to ∞ yields
∞
[zT (t)z(t) − γ 2 f T (t)f (t)]dt < 0 (9.51)
0
which indicates the H∞ performance (9.20) is guaranteed.

In Proposition 9.1, a sufficient condition of the asymptotic stability and the given
wave attenuation level γ is obtained for the closed-loop system (9.17). It can be
seen that the sufficient condition is nonlinear due to the term P BK. To obtain the
gain matrix K in (9.16), denote
Ψ1 = diag{P −1 , P −1 } (9.52)
Ψ2 = diag{P −1 , P −1 , P −1 , P −1 , P −1 , I, I, I, I, I, I, I } (9.53)
Pre- and post-multiply (9.21) by Ψ1 , (9.22) by Ψ2 , and their transposes, respectively;

and set
1
P̄ = P −1 , K̄ = KP −1 , σ̄ =
σ
S̄ = P −1 S1 P −1 , Q̄1 = P −1 Q1 P −1 , Q̄2 = P −1 Q2 P −1
Ω̄ = P −1 ΩP −1 , R̄1 = P −1 R1 P −1 , R̄2 = P −1 R2 P −1
Then, we have the following equivalent version of Proposition 9.1.

Proposition 9.2 For given scalars τm and τM satisfying 0 ≤ τm ≤ τM , r̄ > 0,
γ > 0 and h > 0, if there exist 4 × 4 matrices P̄ > 0, Q̄1 > 0, Q̄2 > 0, R̄1 > 0,
R̄2 > 0, S̄, Ω̄ > 0, a 1 × 4 matrix K̄ and a positive scalar ε such that

R̄2 S̄
≥0 (9.54)
∗ R̄2
⎡ ⎤
Δ̄11 Δ̄12 Δ̄13
⎣ ∗ Δ̄22 Δ23 ⎦ < 0 (9.55)
∗ ∗ Δ33
where Δ23 and Δ33 are given by (9.26) and (9.27), respectively, and
⎡ ⎤
Θ̄ r̄B K̄ R̄1 0 r̄B K̄ D
⎢ ∗ −2R̄ + S̄ + S̄ T R̄2 − S̄ R̄2 − S̄ T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q̄2 − Q̄1 − R̄1 − R̄2 S̄ T 0 0 ⎥
Δ̄11 =⎢ ⎥
⎢∗ ∗ ∗ −Q̄2 − R̄2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω̄ 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
(9.56)
⎡ ⎤ ⎡ ⎤
0 τm P̄ AT (η̄ − τm )P̄ AT P̄ C1T εB 0
⎢ P̄ r̄τ K̄ T B T r̄(η̄ − τ )K̄ T B T 0 ⎥ ⎢ 0 r̄ K̄ T ⎥
⎢ m m ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
Δ̄12 =⎢ ⎥ , Δ̄13 = ⎢ ⎥ (9.57)
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ P̄ r̄τm K̄ T B T r̄(η̄ − τm )K̄ T B T 0 ⎦ ⎣ 0 r̄ K̄ T ⎦
0 τm D T (η̄ − τm )D T D1T 0 0
Δ̄22 = diag{−σ̄ P̄ Ω̄ −1 P̄ , −P̄ R̄1−1 P̄ , P̄ R̄2−1 P̄ , −I } (9.58)
with
Θ̄ = AP̄ + P̄ AT + Q̄1 − R̄1 (9.59)

Then, the closed-loop system (9.17) with f (t) = 0 is asymptotically stable, and the
H∞ performance (9.20) is guaranteed for the wave force f (t) ∈ L2 [0, ∞] and the
prescribed γ . Moreover, the matrix K in (9.16) is given by K = K̄ P̄ −1 .
Due to the nonlinear terms such as σ̄ P̄ Ω̄ −1 P̄ , P̄ R̄1−1 P̄ and P̄ R̄2−1 P̄ in (9.58), it
is difficult to compute the variables P̄ , Q̄1 , Q̄2 , R̄1 , R̄2 , S̄, Ω̄, K̄, and ε and thereby
to obtain the gain matrix K. Noting the fact that
− σ̄ P̄ Ω̄ −1 P̄ ≤ σ̄ Ω̄ − 2σ̄ P̄
− P̄ R̄1−1 P̄ ≤ R̄1 − 2P̄ , −P̄ R̄2−1 P̄ ≤ R̄2 − 2P̄
In this case, we have the following proposition.

Proposition 9.3 For given scalars τm and τM satisfying 0 ≤ τm ≤ τM , r̄ > 0,
γ > 0 and h > 0, if there exist 4 × 4 matrices P̄ > 0, Q̄1 > 0, Q̄2 > 0, R̄1 > 0,
R̄2 > 0, S̄, Ω̄ > 0, a 1 × 4 matrix K̄ and a positive scalar ε such that (9.54) and
⎡ ⎤
Δ̄11 Δ̄12 Δ̄13
⎣ ∗ Δ̂22 Δ23 ⎦ < 0 (9.60)
∗ ∗ Δ33
where Δ̄11 , Δ̄12 , and Δ̄13 are defined in (9.56) and (9.57), Δ23 and Δ33 are given
by (9.26) and (9.27), respectively, and
Δ̂22 = diag{σ̄ Ω̄ − 2σ̄ P̄ , R̄1 − 2P̄ , R̄2 − 2P̄ , −I } (9.61)
then, the closed-loop system (9.17) with f (t) = 0 is asymptotically stable, and the
H∞ performance (9.20) is guaranteed for the wave force f (t) ∈ L2 [0, ∞] and the
prescribed γ . Moreover, the matrix K in (9.16) is given by K = K̄ P̄ −1 .
If there is no any fault in the actuator, one yields an event-triggered H∞ controller
as
u(t) = K(δ(t) + x(t − ρ(t))), t ∈ [tk , tk+1 ) (9.62)
In this situation, the matrix K can be solved via the following corollary, which is a
special case of Proposition 9.3.
Corollary 9.1 For given scalars τm and τM satisfying 0 ≤ τm ≤ τM , γ > 0 and
h > 0, if there exist 4 × 4 matrices P̄ > 0, Q̄1 > 0, Q̄2 > 0, R̄1 > 0, R̄2 > 0, S̄,
Ω̄ > 0, a 1 × 4 matrix K̄ and a positive scalar ε such that (9.54) and

Υ11 Υ12
<0 (9.63)
∗ Δ̂22
where Δ̂22 is defined by (9.61), and

⎡ ⎤
Θ̄ B K̄ R̄1 0 B K̄ D
⎢ ∗ −2R̄ + S̄ + S̄ T R̄2 − S̄ R̄2 − S̄ T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q̄2 − Q̄1 − R̄1 − R̄2 S̄ T 0 0 ⎥
Υ11 =⎢ ⎥
⎢∗ ∗ ∗ −Q̄2 − R̄2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω̄ 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
⎡ ⎤
0 τm P̄ AT (η̄ − τm )P̄ AT P̄ C1T
⎢ P̄ τ K̄ T B T (η̄ − τ )K̄ T B T 0 ⎥
⎢ m m ⎥
⎢ ⎥
⎢0 0 0 0 ⎥
Υ12 =⎢ ⎥ (9.64)
⎢0 0 0 0 ⎥
⎢ ⎥
⎣ P̄ τm K̄ T B T (η̄ − τm )K̄ T B T 0 ⎦
0 τm D T (η̄ − τm )D T D1T
where the matrix Θ̄ is defined by (9.59). Then, under the controller (9.62), the system
(2.3) with f (t) = 0 is asymptotically stable, and the H∞ performance (9.20) is
guaranteed for the wave force f (t) ∈ L2 [0, ∞] and the prescribed γ . Moreover,
the gain matrix K in (9.62) is determined by K = K̄ P̄ −1 .
It should be pointed out that the above results are based on the dynamic model of
the offshore structure, where the first dominant vibration mode is considered. Note
that other higher vibration modes may have some effects on the performance and
stability of the offshore structure. If we take the effects of higher vibration modes
on the first mode and the AMD into account and treat them as the unknown but
bounded nonlinear perturbations, then the corresponding dynamic equations of the
offshore structure can be described by
⎧
⎪
⎪ m1 z̈1 (t) = −c1 ż1 (t) − k1 z1 (t) + k2 (z2 (t) − z1 (t)) + c2 (ż2 (t) − ż1 (t))
⎨
+f (t) − u(t) + g1 (t, z1 (t), z2 (t), ż1 (t), ż2 (t))
⎪
⎪ m z̈ (t) = −c2 (ż2 (t) − ż1 (t)) − k2 (z2 (t) − z1 (t))
⎩ 2 2
+g2 (t, z1 (t), z2 (t), ż1 (t), ż2 (t))
(9.65)
where m1 , m2 , k1 , k2 , c1 , c2 , z1 (t), z2 (t), u(t), and f (t) are defined in the system
(2.1) and g1 (t, z1 , z2 , ż1 , ż2 ) and g2 (t, z1 , z2 , ż1 , ż2 ) present the effects of higher
vibration modes on the first mode and the AMD, respectively.
It is supposed that the perturbation terms g1 (t, z1 , z2 , ż1 , ż2 ) and g2 (t, z1 , z2 ,
ż1 , ż2 ) satisfy
T
giT (t, z1 , z2 , ż1 , ż2 )gi (t, z1 , z2 , ż1 , ż2 ) ≤ ςi2 z1 z2 ż1 ż2 z1 z2 ż1 ż2
gi (t, 0, 0, 0, 0) = 0, i = 1, 2
(9.66)
where ςi > 0, i = 1, 2.
Define
⎡ 1 ⎤T
00 0
⎢ m ⎥ g1 (t, x(t))
D0 = ⎣ 1
1 ⎦ , g(t, x(t)) = g2 (t, x(t)) (9.67)
00 0
m2
Then, the Eq. (9.65) can be written as
ẋ(t) = Ax(t) + Bu(t) + Df (t) + D0 g(t, x(t)), x(0) = x0 (9.68)
where the matrices A, B and D are given by (2.4).

It is clear from (9.66) and (9.67) that the nonlinear term g(t, x(t)) satisfies the
following constraint:

g T (t, x(t))g(t, x(t)) ≤ ς 2 x T (t)x(t)
(9.69)
g(t, 0) = 0
where ς = ς12 + ς22 .

In this case, under the event-triggered H∞ reliable controller (9.16), to make
the system (9.68) with f (t) = 0 asymptotically stable, and guarantee the H∞
performance (9.20), one can compute the gain matrix K of the controller (9.16)
by solving the linear matrix inequalities (9.54) and
⎡ ⎤
Δ̂11 Δ̂12 Δ̂13
⎣ ∗ Δ̂22 Δ̂23 ⎦ < 0 (9.70)
∗ ∗ Δ̂33
where Δ̂22 is given by (9.61), and

⎡ ⎤
Θ̄ r̄B K̄ R̄1 0 r̄B K̄ D D0
⎢ ∗ −2R̄ + S̄ + S̄ T R̄2 − S̄ R̄2 − S̄ T 0 ⎥
⎢ 2 0 0 ⎥
⎢∗ ∗ Q̄2 − Q̄1 − R̄1 − R̄2 S̄ T 0 ⎥
⎢ 0 0 ⎥
⎢ ⎥
Δ̂11 =⎢∗ ∗ ∗ −Q̄2 − R̄2 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ ∗ −Ω̄ 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −γ 2 I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −I
(9.71)
⎡ ⎤ ⎡ ⎤
0 τm P̄ AT (η̄ − τm )P̄ AT P̄ C1T εB 0 ς P̄
⎢ P̄ r̄τ K̄ T B T r̄(η̄ − τ )K̄ T B T 0 ⎥ ⎢ 0 r̄ K̄ T 0 ⎥
⎢ m m ⎥ ⎢ ⎥
⎢0 0 ⎥ ⎢ 0 0 0 ⎥
⎢ 0 0 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
Δ̂12 =⎢0 0 0 0 ⎥ , Δ̂13 = ⎢ 0 0 0 ⎥
⎢ ⎥ ⎢ ⎥
⎢ P̄ r̄τm K̄ T B T r̄(η̄ − τm )K̄ T B T 0 ⎥ ⎢ 0 r̄ K̄ T 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 τm D T (η̄ − τm )D T D1T ⎦ ⎣ 0 0 0 ⎦
0 τm D0 T (η̄ − τm )D0 T 0 0 0 0
(9.72)
⎡ ⎤T
0 ετm B T ε(η̄ − τm )B T 0
Δ̂23 = ⎣0 0 0 0⎦ (9.73)
0 0 0 0
Δ̂33 = diag{−εI, −εI, −I } (9.74)
where h > 0, γ > 0, ς > 0, τm , and τM are given scalars satisfying 0 ≤ τm ≤ τM ,

the matrix Θ̄ is given by (9.59), P̄ > 0, Q̄1 > 0, Q̄2 > 0, R̄1 > 0, R̄2 > 0, S̄,
Ω̄ > 0, K̄, and ε > 0 are matrix variables with appropriate dimensions. If the linear
matrix inequalities (9.54) and (9.70) have feasible solutions, then one can yields the
gain matrix as K = K̄ P̄ −1 .
In this section, the related parameters of an offshore structure are first given. Then,
event-triggered H∞ controllers are designed, respectively, for the system without
actuator faults and the system with actuator faults. Under the obtained controllers,
the performance of the structure is investigated.
Suppose that the parameters of the offshore platform [79] is set as Table 3.1.
Based on the setting in the table, one yields the matrices A and B in system (2.3)
as (3.26). Choose displacements of the offshore structure and the AMD as the
controlled outputs, in this case, the matrices C1 and D1 in (9.1) are given as

1000 0.01
C1 = , D1 =
0010 0
Based on (2.18), the irregular wave force can be computed and presented in
Fig. 9.2. In what follows, we focus on designing event-triggered H∞ controllers,
respectively, for the system without actuator faults and the system with actuator
faults. We set the H∞ performance index γ as 0.1 and the sampling period h as 0.01
second (s). On the one hand, to demonstrate the effectiveness of the proposed control
schemes, the controlled vibration amplitudes of the displacement and acceleration
of the offshore structure are computed, respectively; on the other hand, to measure
5
x 10
5
1
Control Force (N)
−1
−2
−3
−4
−5
0 10 20 30 40 50 60
Time (s)
Fig. 9.2 Wave force acting on the offshore structure
the performance of the event-triggered scheme, the transmission rate (TR) of the
packets [100] is investigated.
It can be obtained that the maximum values of the displacement and acceleration
of the structure without control are 0.2829 m and 0.7987 m/s2 , respectively, and the
root mean square (RMS) values of them are 0.1025 m and 0.3088 m/s2 , respectively.
9.3.1 Event-Triggered H∞ Control
In this subsection, an event-triggered H∞ controller (ETHC) is designed and applied

to the offshore structure without any actuator faults. Let the network-induced delay
change from τm = 0.04 s to τM = 0.08 s randomly. In the event-triggering
condition (9.2), set σ = 0.40. Thus, by Corollary 9.1, one yields the event-triggering
parameter Ω and the gain matrix K as:
⎡ ⎤
1.2529 −0.0064 −0.0602 −0.0187
⎢ −0.0064 0.0006 0.0095 0.0002 ⎥
Ω = 103 × ⎢
⎣ −0.0602 0.0095 0.1630 0.0023 ⎦
⎥
−0.0187 0.0002 0.0023 0.0003

K = 106 × 2.6890 −0.0106 −0.0757 −0.0397
Under the obtained ETHC, the maximum values of the displacement and
acceleration of the offshore structure are reduced from 0.2829 m and 0.7987 m/s2
to 0.2221 m and 0.4627 m/s2 , respectively; the RMS values of them are reduced
from 0.1025 m and 0.3088 m/s2 to 0.0859 m and 0.1934 m/s2 , respectively. One
can conclude that under the designed controller, the vibration amplitudes of the
displacement and acceleration of the structure are effectively reduced. The response
curves of the offshore structure without control and with the ETHC are presented in
Figs. 9.3 and 9.4. To suppress the vibration to such levels, the peak and RMS values
of the required force are 4.7759 ×105 N and 1.9171 ×105 N, respectively. Depicted
in Fig. 9.5 is the curve of the control force by the ETHC, and in Fig. 9.6 is the
diagram of the release time intervals. In fact, one can obtain that under the ETHC,
the TR on [0, 60 s] is 0.0247, which means that the performance of the structure can
be maintained while 97.53% network resources are saved.
To compare the event-triggered H∞ control scheme with the traditional H∞ con-
trol scheme without network setting, we require to study the vibration amplitudes of
the structure with the H∞ control scheme. For this purpose, set the H∞ performance
index γ as 0.1, which is equal to the one for the ETHC. Thus, by Corollary 9.1 in
[89], the gain matrix of an H∞ controller (HIC) can be obtained as

K = 106 × 4.4268 0.1639 0.9081 −0.0709
Under this controller, the peak and RMS values of the structure responses and
the ranges of the required force are given by Table 9.1, where Md , Ma , and
Mu denote the peak values of displacement, acceleration of the offshore platform,
0.3
No Control
Event−triggered H ∞ Controller
0.2
Displacement of Offshore Platform (m)
0.1
−0.1
−0.2
−0.3
0 10 20 30 40 50 60
Time (s)
Fig. 9.3 The displacement responses of system under no control and ETHC
0.8
No Control
Event−triggered H ∞ Controller
0.6
0.4
Acceleration of Offshore Platform (m/s )
2
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60
Time (s)
Fig. 9.4 The acceleration responses of system under no control and ETHC
5
x 10
5
1
Control Force (N)
−1
−2
−3
−4
−5
0 10 20 30 40 50 60
Time (s)
Fig. 9.5 The control force required by ETHC
and the required control force, respectively; Jd , Ja , and Ju represent the RMS
values of displacement, acceleration of the offshore platform, and the control force,
respectively (3.27). The table indicates that both the maximum and the RMS values
3.5
3
Event−based Release Instants and Release Interval
2.5
1.5
0.5
0
0 10 20 30 40 50 60
Time (s)
Fig. 9.6 Release instants and time intervals under ETHC
Table 9.1 The peak and RMS values of displacement, acceleration of the offshore structure, and
the range of the control force under ETHC and HIC
Controllers (s) Md (m) Ma (m/s) Mu (105 N) Jd (m) Ja (m/s) Ju (105 N)
No control 0.2829 0.7987 – 0.1025 0.3088 –
HIC [89] 0.2171 0.4188 14.122 0.0841 0.1760 5.7285
ETHC 0.2221 0.4627 4.7759 0.0859 0.1934 1.9171
of the vibration amplitudes of the structure under the HIC are slightly smaller than
those under the ETHC. However, it is not difficult to see that the required force by
the HIC is larger than the one by the ETHC. In fact, the maximum and the RMS
values of the force by the HIC are nearly three times as those by the ETHC.
Remark 9.2 Compared with the proposed event-triggered H∞ control scheme,
some existing control schemes without network environments, such as the H2
control [79], the feedforward and feedback optimal control [81, 82], the optimal
tracking control [84], and the delayed H∞ control [89], may provide smaller
vibration amplitudes of the structure. However, the required control force by the
proposed event-triggered H∞ control scheme is much less than the one by some
existing control schemes without network environments. Specifically, notice that
the designed ETHC is of several advantages of the network-based control as well
as saving communication resources. Therefore, choosing either the event-triggered
H∞ control scheme or other control schemes without network setting is a tradeoff
between the vibration amplitudes of the structure and the required control force.
In the next subsection, in the case of the offshore structure with actuator
faults, the performance of the system under event-triggered H∞ reliable controller
(ETHRC) is investigated.
9.3.2 Event-Triggered H∞ Reliable Control: Constant Delays
Set the event-triggering parameter σ = 0.16. It is supposed that the network-

induced delays are time-invariant, here τk ≡ 0.03 s, k = 1, 2, · · · and the actuator
faults appear on the time interval [20, 40 s] with the periodical form expressed by
r(t) = 0.25sin(8t). Thus, by Proposition 9.3, the event-triggering parameter Ω and
the gain matrix K are obtained as:
⎡ ⎤
1.9804 −0.0048 0.0056 −0.0293
⎢ −0.0048 0.0017 0.0281 0.0003 ⎥
Ω = 103 × ⎢
⎣ 0.0056 0.0281 0.4663 0.0039 ⎦
⎥
−0.0293 0.0003 0.0039 0.0005

K = 106 × 2.1644 −0.0021 0.0581 −0.0316
The event-triggered H∞ reliable controller is denoted by ETHRC1. Presented by

Figs. 9.7 and 9.8 are response curves of the structure under the ETHRC1, and
Fig. 9.9 is the control force curve. In this situation, one yields that the maximum
0.3
No Control
Event−triggered H ∞ Reliable Controller
0.2
0.1
−0.1
−0.2
0 10 20 30 40 50 60
Time (s)
Fig. 9.7 The displacement responses of system under no control and ETHRC1
0.8
No Control
0.6
0.4
2
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60
Time (s)
Fig. 9.8 The acceleration responses of system under no control and ETHRC1
5
x 10
5
2
Control Force (N)
−1
−2
−3
−4
0 10 20 30 40 50 60
Time (s)
Fig. 9.9 The control force required by ETHRC1
vibration amplitudes of the displacement and acceleration of the system as 0.2268 m

and 0.5211 m/s2 , respectively, and the RMS values of them as 0.0917 m and
0.2369 m/s2 , respectively. Correspondingly, the maximum and the RMS values of
1.6
1.4
1.2
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60
Time (s)
Fig. 9.10 Release instants and time intervals under ETHRC1
the required force are about 4.1065 ×105 N and 1.4334 ×105 N, respectively. It is
found that if there exist actuator faults in the offshore structure, the performance of
the structure can be ensured. In addition, the TR for the ETHRC1 is 0.0478, which
indicates that during the control process, only 4.78% communication resources
are transmitted to the ZOH through the communication network, while 97.53%
resources are saved. Figure 9.10 presents a diagram of release time intervals versus
release instants.
9.3.3 Event-Triggered H∞ Reliable Control: Time-Varying

Delays
Suppose that the network-induced delays vary from τm = 0.02 to τM = 0.04 s

randomly; the event-triggering parameter σ is taken as 0.156, an actuator fault of
the offshore structure occurs on time interval [20, 40 s]. The fault is simulated by a
random signal ranging between 0.05 and 0.15. Then, applying Proposition 9.3 yields
the event-triggering parameter Ω and the gain matrix K as:
⎡ ⎤
1.5299 −0.0037 0.0044 −0.0226
⎢ −0.0037 0.0013 0.0217 0.0002 ⎥
Ω = 10 × ⎢
3
⎣ 0.0044
⎥
0.0217 0.3602 0.0030 ⎦
−0.0226 0.0002 0.0030 0.0004

K = 106 × 3.4059 −0.0032 0.0939 −0.0497
Denote the obtained controller as ETHRC2. When this controller is utilized

to control the offshore structure, we can compute that the offshore structure
vibrates with the maximum oscillation amplitudes of 0.2146 m for displacement and
0.5285 m/s2 for acceleration, respectively. The RMS values of the displacement and
acceleration are reduced to 0.0916 m and 0.2337 m/s2 , respectively. The maximum
and RMS values of the required force are 5.8374 ×105 N and 2.1052 ×105 N,
respectively. It is clear that the controller can still stabilize the structure subject to
the external wave force as well as actuator faults, which can be seen from Figs. 9.11,
9.12, and 9.13. In this case, the release time intervals are demonstrated by Fig. 9.14.
In fact, it is readily obtained that the value of TR for ETHRC2 is 0.0455, which
shows that proposed controller is capable of reducing the communication resources
of the network significantly.
To investigate the effects of the network-induced delays on the performance of
the structure, when the network-induced delays increase gradually, the maximum
and RMS values of vibration amplitudes of the offshore structure under ETHC,
ETHRC1, and ETHRC2 are listed in Tables 9.2, 9.3, and 9.4, respectively. The
Tables show that the vibration amplitudes of the structure, the force, and the
transmission rate of the sampled-data packets become large with the increase of
the network-induced delays. In fact, under these controllers, if the network-induced
delays vary in a proper range, such as (0, 0.7 s] for ETHRC1, the communication
resources are greatly saved, meanwhile, the vibrations of the displacement and the
acceleration of the structure are attenuated effectively.
0.3
No Control
0.2
0.1
−0.1
−0.2
−0.3
0 10 20 30 40 50 60
Time (s)
Fig. 9.11 The displacement responses of system under no control and ETHRC2
0.8
No Control
Event−triggered H Reliable Controller
∞
0.6
0.4
2
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60
Time (s)
Fig. 9.12 The acceleration responses of system under no control and ETHRC2
5
x 10
6
2
Control Force (N)
−2
−4
−6
0 10 20 30 40 50 60
Time (s)
Fig. 9.13 The control force required by ETHRC2
Based on the above simulation results, one can see that the following statements
are true:
• If there is no any actuator fault in the control process, the designed event-
triggered H∞ controller can effectively attenuate the vibration of the offshore
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Time (s)
Fig. 9.14 Release instants and time intervals under ETHRC2
Table 9.2 The peak and RMS values of displacement, acceleration of the offshore structure, the
control force, and the TR under ETHC with τm = 0.02 s and τM varies
τM (s) Md (m) Ma (m/s) Mu (105 N) Jd (m) Ja (m/s) Ju (105 N) TR (%)
0.03 0.2213 0.4601 4.8412 0.0860 0.1926 1.9567 2.52
0.07 0.2215 0.4601 4.8436 0.0859 0.1927 1.9496 2.48
0.15 0.2221 0.4574 4.8619 0.0856 0.1928 1.9381 2.50
0.25 0.2231 0.4584 4.8530 0.0854 0.1943 1.9290 2.58
0.40 0.2240 0.4701 4.7654 0.0850 0.1946 1.9223 2.63
0.70 0.2366 0.5325 5.2742 0.0868 0.2228 1.9964 2.87
0.90 0.2306 0.6739 5.4838 0.0917 0.2676 2.2753 3.42
1.00 0.2645 0.6659 6.3192 0.0959 0.2991 2.6886 3.47
1.01 0.2924 0.8535 8.1240 0.0960 0.3072 2.5185 3.78
structure, and reduce the control cost dramatically. If there exist actuator faults
in the system, the event-triggered H∞ reliable controller can also stabilize the
platform and thus improve performance of the offshore structure.
• Under the event-triggered H∞ controllers, the network resources can be saved
significantly, thereby a good network service can be guaranteed and satisfactory
performance of the offshore platform can be ensured.
control force, and the TR under ETHRC1 for different values of constant network-induced delay τ
0.02 0.2268 0.5211 4.1062 0.0917 0.2367 1.4359 4.70
0.08 0.2273 0.5210 4.1132 0.0916 0.2378 1.4228 4.72
0.14 0.2279 0.5219 4.1265 0.0916 0.2394 1.4117 4.70
0.25 0.2295 0.5318 4.1711 0.0918 0.2437 1.3975 4.83
0.40 0.2329 0.5647 4.2919 0.0927 0.2539 1.3969 4.77
0.60 0.2416 0.6565 4.6376 0.0960 0.2850 1.4763 4.92
0.70 0.2498 0.7404 4.9565 0.1006 0.3213 1.6141 5.25
control force, and the TR under ETHRC2 with τm = 0.01 s, and the values of τM are different τ
0.03 0.2146 0.5287 5.8426 0.0916 0.2340 2.1104 4.48
0.12 0.2150 0.5338 5.8346 0.0915 0.2349 2.0827 4.55
0.25 0.2159 0.5443 5.8184 0.0915 0.2377 2.0518 4.88
0.40 0.2170 0.5509 5.9327 0.0918 0.2426 2.0429 4.67
0.57 0.2183 0.5627 6.1172 0.0917 0.2462 2.0611 4.73
0.65 0.2241 0.5919 6.1092 0.0923 0.2548 2.0888 4.75
0.75 0.2236 0.6391 6.8572 0.0951 0.2784 2.1667 5.35
0.80 0.2344 0.7243 7.5623 0.0977 0.3018 2.4874 6.13
9.4 Conclusions
In this chapter, an event-triggered H∞ reliable control problem for an offshore

platform under external wave force and actuator faults has been investigated. An
event-triggering approach has been presented to deal with the issue on the limited
network bandwidth in the controller design. By applying the stability criterion, the
sufficient conditions have been provided for the existence of event-triggered H∞
reliable controller. Simulation results have been given to illustrate the validity of the
proposed method.
9.5 Notes
This chapter is mainly based on Zhang and Han [123]. Another result of network-
based controller design of offshore platform is reported in [120]. In fact, in [120],
a design approach of network-based state feedback controller for an offshore steel
jacket platform subject to self-excited nonlinear hydrodynamic force is provided,
9.5 Notes 181
and the effects of network-induced delay on the active control for the platform is
investigated. However, the main concern of [123] is that, based on an offshore steel
jacket platform subject to external wave force, to deal with the limited resources
of communication networks and actuator faults [85]. As is known that unexpected
actuator faults and fatigue damage often appear in the implementation of the
offshore platform systems. Such faults and even fatigue damage generally lead to
poor performance and even instability of the platforms. To prevent fatigue damage
of offshore platforms and to protect operation and staff on offshore platforms subject
to a wide range of environmental loading, it is very important for offshore platforms
to improve the ability to detect, diagnose, and tolerate malfunctions of the control
system. To develop reliable fault diagnosis and fault-tolerant controllers for the
offshore platforms under network settings to cope with potential failures in actuators
and sensors is one of significant issues.
Another issue is that vibration control of offshore platforms lies at an inter-
section of different research areas including structure vibration theory, control
theory, communication theory, civil engineering, mechanical engineering, and ocean
engineering. Consequently, collaborative research in vibration control of offshore
platforms is required from different engineering and scientific fields.
Compared with passive and semi-active control, active control has several
distinct advantages. However, if power supply cannot be guaranteed during the harsh
ocean environment, the active controller will not work [4]. In this situation, semi-
active and hybrid control mechanisms with passive components are still feasible
options for the vibration control of the offshore platforms. Therefore, to develop
semi-active and hybrid control mechanisms with high reliability, desirable control
effects and low control cost requires further study.
References
1. Wilson, J.F. (ed.): Dynamics of Offshore Structures. Wiley, Chichester (2002)

2. Hirdaris, S.E., Bai, W., Dessi, D., et al.: Loads for use in the design of ships and offshore
structures. Ocean Eng. 78, 131–174 (2014)
3. Ou, J., Long, X., Li, Q.S., et al.: Vibration control of steel jacket offshore platform structures
with damping isolation systems. Eng. Struct. 29(7), 1525–1538 (2007)
4. Kandasamy, R., Cui, F., Townsend, N., et al.: A review of vibration control methods for marine
offshore structures. Ocean Eng. 127, 279–297 (2016)
5. Zhang, B.-L., Han, Q.-L., Zhang, X.-M.: Recent advances in vibration control of offshore
platforms. Nonlinear Dyn. 89(2), 755–771 (2017)
6. Soong, T.T., Dargush, G.F. (eds.): Passive Energy Dissipation Systems in Structural Engineer-
ing. Wiley, Buffalo (1997)
7. Yao, J.T.P.: Concept of structural control. J. Struct. Div. 98(7), 1567–1574 (1972)
8. Korkmaz, S.: A review of active structural control: challenges for engineering informatics.
Comput. Struct. 89(23–24), 2113–2132 (2011)
9. Lee, H.H.: Stochastic analysis for offshore structures with added mechanical dampers. Ocean
Eng. 24(5), 817–834 (1997)
10. Li, H., Wang, S., Ji, C.: Semi-active control of wave-induced vibration for offshore platforms
by use of MR damper. China Ocean Eng. 16(1), 33–40 (2002)
11. Abdel-Rohman, M.: Structural control of a steel jacket platform. Struct. Eng. Mech. 4(2),
125–138 (1996)
12. Li, H.-N., He, X.-Y., Huo, L.-S.: Seismic response control of offshore platform structures with
shape memory alloy Dampers. China Ocean Eng. 19(2), 185–194 (2005)
13. Patil, K.C., Jangid, R.S.: Passive control of offshore jacket platforms. Ocean Eng. 32, 1933–
1949 (2005)
14. Golafshani, A.A., Gholizad, A.: Friction damper for vibration control in offshore steel jacket
platforms. J. Constr. Steel Res. J. Constr. Steel Res. 65(1), 180–187 (2009)
15. Moharrami, M., Tootkaboni, M.: Reducing response of offshore platforms to wave loads using
hydrodynamic buoyant mass dampers. Eng. Struct. 81, 162–174 (2014)
16. Monir, H.S., Nomani, H.: Application of lead rubber isolation systems in the offshore struc-
tures. In: Proceedings of the International MultiConference of Engineering and Computer
Scientists, Hong Kong, pp. 1523–1527 (2011)
17. Liu, X., Li, G., Yue, Q., et al.: Acceleration-oriented design optimization of ice-resistant jacket
platforms in the Bohai Gulf. Ocean Eng. 36(17–18), 1295–1302 (2009)
18. Wang, S., Yue, Q., Zhang, D.: Ice-induced non-structure vibration reduction of jacket
platforms with isolation cone system. Ocean Eng. 70(15), 118–123 (2013)

https://doi.org/10.1007/978-981-13-2986-9
184 References
19. Kareem, A.: Mitigation of wind induced motion of tall buildings. J. Wind Eng. Ind. Aerod.
11(1–3), 273–284 (1983)
20. Alves, R.M., Batista, R.C.: Active/passive control of heave motion for TLP type of offshore
platforms. In: Proceedings of the International Offshore and Polar Engineering Conference,
Brest, France, pp. 332–338 (1999)
21. Wang, S., Li, H., Ji, C., et al.: Energy analysis for TMD-structure systems subjected to impact
loading. China Ocean Eng. 16(3), 301–310 (2002)
22. Chandrasekaran, S., Bhaskar K., Lino, H., et al.: Dynamic response behaviour of multi-legged
articulated tower with & without TMD. In: Proceedings of the International Conference on
Marine Technology, Dhaka, Bangladesh. pp. 131–136 (2010)
23. Yue, Q., Zhang, L., Zhang, W., et al.: Mitigating ice-induced jacket platform vibrations
utilizing a TMD system. Cold Reg. Sci. Technol. 56(2–3), 84–89 (2009)
24. Abe, M., Igusa, T.: Tuned mass dampers for structures with closely spaced natural frequen-
cies. Earthq. Eng. Struct. Dyn. 24, 247–261 (1995)
25. Taflanidis, A.A., Angelides, D.C., Scruggs, J.T.: Robust design optimization of mass dampers
for control of tension leg platforms. In: Proceedings of the International Offshore and Polar
Engineering Conference, Vancouver, Canada, pp. 92–99 (2008)
26. Lu, J., Mei, N., Li, Y., et al.: Vibration control of multi-tuned mass dampers for an offshore
oil platfrom. China Ocean Eng. 16(3), 321–328 (2002)
27. Taflanidis, A.A., Angelides, D.C., Scruggs, J.T.: Simulation-based robust design of mass
dampers for response mitigation of tension leg platforms. Eng. Struct. 31(4), 847–857 (2009)
28. Chandrasekaran, S., Kumar, D., Ramanathan, R.: Dynamic response of tension leg platform
with tuned mass dampers. J. Naval Architect. Mar. Eng. 10(2), 1813–8235 (2013)
29. Ma, R., Wang, J., Zhao, D.: Simulation of vibration control of offshore platforms under
earthquake loadings. In: Proceedings of the International Conference on Offshore Mechanics
and Arctic Engineering, Berlin, Germany, pp. 597–601 (2008)
30. Zhao, D., Cai, D.M., Ma, R.J.: Vibration control of offshore platforms using METMD system
under the random ocean wave forces. In: Proceedings of the Seventh ISOPE Pacific/Asia
Offshore Mechanics Symposium, Dalian, China, pp. 60–65 (2006)
31. Golafshani, A.A., Gholizad, A.: Passive devices for wave induced vibration control in
offshore steel jacket platforms. Trans. A Civ. Eng. 16(6), 443–456 (2009)
32. Jafarabad, A., Kashani, M., Parvar, M.R.A., et al.: Hybrid damping systems in offshore jacket
platforms with float-over deck. J. Constr. Steel Res. 98, 178–187 (2014)
33. Ma, R., Zhang, H., Zhao, D.: Study on the anti-vibration devices for a model jacket platform.
Mar. Struct. 23(4), 434–443 (2010)
34. Vel.ičko, J., Gaile L.: Overview of tuned liquid dampers and possible ways of oscillation
damping properties improvement. In: Proceedings of the International Scientific and Practical
Conference on Environment, Technology, Resources, Rezekne, Latvia, pp. 233–238 (2015)
35. Vandiver, J.K., Mitome, S.: Effect of liquid storage tanks on the dynamic response of offshore
platforms. Appl. Ocean Res. 1(2), 67–74 (1979)
36. Li, H., Ma, B.: Seismic response reduction for fixed offshore platform by tuned liquid damper.
China Ocean Eng. 11(2), 119–125 (1997)
37. Chen, X., Wang, L., Xu, J.: TLD technique for reducing ice-induced vibration on platforms.
J. Cold Reg. Eng. 13(3), 139–152 (1999)
38. Jin, Q., Li, X., Sun, N., et al.: Experimental and numerical study on tuned liquid dampers
for controlling earthquake response of jacket offshore platform. Mar. Struct. 20(4), 238–254
(2007)
39. Spillane, M.W., Rijken, O.R., Leverette S.J.: Vibration absorbers for deep water TLP’s.
In: Proceedings of the International Offshore and Polar Engineering Conference, Lisbon,
Portugal, pp. 210–217 (2007)
40. Bian, X.S., Leverette, S.J., Rijken, O.R.: A TLP solution for 8000 ft water depth. In:
Proceedings of the International Conference on Ocean, Offshore and Arctic Engineering,
Shanghai, China, pp. 255–262 (2010)
References 185
41. Rijken, O., Spillane, M., Leverette, S.J.: Vibration absorber technology and conceptual
design of vibration absorber for TLP in ultradeep water. In: Proceedings of the International
Conference on Ocean, Offshore and Arctic Engineering, Shanghai, China, pp. 629–638
(2010)
42. Sakai, F., Takaeda, S., Tamaki, T.: Tuned liquid column damper-new type device for
suppression of building vibrations. In: Proceedings of the International Conference on
Highrise Buildings, Nanjing, China, pp. 926–931 (1989)
43. Chaiviriyawong, P., Webster, W.C., Pinkaew, T., et al.: Simulation of characteristics of tuned
liquid column damper using a potential-flow method. Eng. Struct. 29(1), 132–144 (2007)
44. Lee, H.H., Wong, S.-H., Lee, R.-S.: Response mitigation on the offshore floating platform
system with tuned liquid column damper. Ocean Eng. 33(8–9), 1118–1142 (2006)
45. Huo, L., Li, H.: Torsionally coupled response control of offshore platform structures using
Circular Tuned Liquid Column Dampers. China Ocean Eng. 18(2), 173–183 (2004)
46. Al-Saif, K.A., Aldakkan, K.A., Foda, M.A.: Modified liquid column damper for vibration
control of structures. Int. J. Mech. Sci. 53(7), 505–512 (2011)
47. Chatterjee, T., Chakraborty, S.: Vibration mitigation of structures subjected to random wave
forces by liquid column dampers. Ocean Eng. 87(1), 151–161 (2014)
48. Lee, H.H., Juang, H.H.: Experimental study on the vibration mitigation of offshore tension
leg platform system with UWTLCD. Smart Struct. Syst. 9(1), 71–104 (2012)
49. Mousavi, S.A., Zahrai, S.M., Bargi, K.: Optimum geometry of tuned liquid column-gas
damper for control of offshore jacket platform vibrations under seismic excitation. Earthq.
Eng. Eng. Vib. 11(4), 579–592 (2012)
50. Mousavi, S.A., Bargi, K., Zahrai, S.M. Optimum parameters of tuned liquid column-gas
damper for mitigation of seismic-induced vibrations of offshore jacket platforms. Struct.
Control. Health Monit. 20(3), 422–444 (2013)
51. Hochrainer, M.J., Ziegler, F.: Control of tall building vibrations by sealed tuned liquid column
dampers. Struct. Control. Health Monit. 13(6), 980–1002 (2006)
52. Ziegler, F.: Special design of tuned liquid column-gas dampers for the control of spatial
structural vibrations. Acta Mech. 201(1), 249–267 (2008)
53. Zeng, X., Yu, Y., Zhang, L., et al.: A new energy-absorbing device for motion suppression in
deep-sea floating platforms. Energies 8(1), 111–132 (2015)
54. Pinkaew, T., Fujino, Y.: Effectiveness of semi-active tuned mass dampers under harmonic
excitation. Eng. Struct. 23(7), 850–856 (2001)
55. Spencer, B.F. Jr, Dyke, S.J., Sain, M.K., et al.: Phenomenological model of a magnetorheo-
logical damper. J. Eng. Mech. 123(3), 230–238 (1997)
56. Karkoub, M., Lamont, L.A., Chaar, L.E.: Design of a test rig for vibration control of oil
platforms using Magneto-Rheological Dampers. J. Offshore Mech. Arct. Eng. 133(4), 041302
(2011). https://doi.org/10.1115/1.4003358
57. Sarrafan, A., Zareh, S.H., Khayyat, A.A., et al.: Performance of an offshore platform with
MR dampers subjected to wave. In: Proceedings of the IEEE International Conference on
Mechatronics, Istanbul, Turkey, pp. 242–247 (2011)
58. Ji, C., Yin, Q.: Study on a fuzzy MR damper vibration control strategy for offshore
platforms. In: Proceedings of the International Conference on Offshore Mechanics and Arctic
Engineering, San Diego, USA, pp. 363–368 (2007)
59. Ji, C., Chen, M., Li S.: Vibration control of jacekt platforms with magnetorheological damper
and experimental validation. High Technol. Lett. 16(2), 189–193 (2010)
60. Wu, B., Shi, P., Wang, Q., et al.: Performance of an offshore platform with MR dampers
subjected to ice and earthquake. Struct. Control Health Monit. 18(6), 682–697 (2011)
61. Wang, S.-Q., Li, N.: Semi-active vibration control for offshore platforms based on LQG
method. J. Mar. Sci. Technol. 21(5), 562–568 (2013)
62. Sarrafan, A., Zareh, S.H., Khayyat, A.A.A., et al.: Neuro-fuzzy control strategy for an
offshore steel jacket platform subjected to wave-induced forces using magnetorheological
dampers. J. Mech. Sci. Technol. 26(4), 1179–1196 (2012)
186 References
63. Taghikhany, T., Ariana, Sh., Mohammadzadeh, R., et al.: The effect of semi-active controller
in Sirri jacket seismic vibration control under Kobe earthquake. Int. J. Mar. Sci. Eng. 3(2),
77–84 (2013)
64. Fischer, F.J., Liapis, S.I., Kallinderis, Y.: Mitigation of current-driven, vortex-induced vibra-
tions of a spar platform via “SMART” thrusters. J. Offshore Mech. Arct. Eng. 126(1), 96–104
(2004)
65. Zribi, M., Almutairi, N., Abdel-Rohman, M., et al.: Nonlinear and robust control schemes for
offshore steel jacket platforms. Nonlinear Dyn. 35(1), 61–80 (2004)
66. Zhang, B.-L., Hu, Y.-H., Tang, G.-Y.: Stabilization control for offshore steel jacket platforms
with actuator time-delays. Nonlinear Dyn. 70(2), 1593–1603 (2012)
67. Suhardjo, J., Kareem, A.: Structural control of offshore platforms. In: Proceedings of the
International Offshore and Polar Engineering Conference, Honolulu, USA, pp. 416–424
(1997)
68. Nakamura, M., Kajiwara, H., Koterayama, W., et al.: Control system design and model
experiments on thruster assisted mooring system. In: Proceedings of the International
Offshore and Polar Engineering Conference, Honolulu, USA, pp. 641–648 (1997)
69. Yamamoto, I., Matsuura, M., Yamaguchi, Y., et al.: Dynamic positioning system based on
nonlinear programming for offshore platforms. In: Proceedings of the International Offshore
and Polar Engineering Conference, Honolulu, USA, pp. 632–640 (1997)
70. Suhardjo, J., Kareem, A.: Feedback-feedforward control of offshore platforms under random
waves. Earthq. Eng. Struct. Dyn. 30, 213–235 (2001)
71. Kawano, K.: Active control effects on dynamic response of offshore structures. In: Proceed-
ings of the International Offshore and Polar Engineering Conference, Singapore, pp. 494–498
(1993)
72. Terro, M.J., Mahmoud, M.S., Abdel-Rohman, M.: Multi-loop feedback control of offshore
steel jacket platforms. Comput. Struct. 70(2), 185–202 (1999)
73. Mahadik, A.S., Jangid, R.S.: Active control of offshore jacket platforms. Int. Shipbuild. Progr.
50(4), 277–295 (2003)
74. Luo, M., Zhu, W.Q.: Nonlinear stochastic optimal control of offshore platforms under wave
loading. J. Sound Vib. 296(4–5), 734–745 (2006)
75. Suneja, B.P., Datta, T.K.: Active control of ALP with improved performance function. Ocean
Eng. 25(10), 817–835 (1998)
76. Yoshida, K., Suzuki, H., Nam, D.: Active control of coupled dynamic response of TLP hull
and tendon. In: Proceedings of the International Offshore and Polar Engineering Conference,
Osaka, Japan, pp. 98–104 (1994)
77. Ahmad, S.K., Ahmad, S.: Active control of non-linearly coupled TLP response under wind
and wave environments. Comput. Struct. 72(6), 735–747 (1999)
78. Alves, R.M., Battista, R.C., Albrecht, C.H.: Active control for enhancing fatigue life of
TLP platform and tethers. In: Proceedings of the International Congress of Mechanical
Engineering, Sao Paulo, Brazil (2003)
79. Li, H.-J., Hu, S.-L., Jakubiak, C.: H2 active vibration control for offshore platform subjected
to wave loading. J. Sound Vib. 263(4), 709–724 (2003)
80. Wang, W., Tang, G.-Y.: Feedback and feedforward optimal control for offshore jacket
platforms. China Ocean Eng. 18(4), 515–526 (2004)
81. Ma, H., Tang, G.-Y., Zhao, Y.-D.: Feedforward and feedback optimal control for offshore
structures subjected to irregular wave forces. Ocean Eng. 33(8–9), 1105–1117 (2006)
82. Ma, H., Tang, G.-Y., Hu, W.: Feedforward and feedback optimal control with memory for
offshore platforms under irregular wave forces. J. Sound Vib. 328(4–5), 369–381 (2009)
83. Zhang, B.-L., Liu, Y.-J., Han, Q.-L., et al.: Optimal tracking control with feedforward
compensation for offshore steel jacket platforms with active mass damper mechanisms. J.
Vib. Control 22(3), 695–709 (2016)
84. Zhang, B.-L., Liu, Y.-J., Ma, H., et al.: Discrete feedforward and feedback optimal tracking
control for offshore steel jacket platforms. Ocean Eng. 91, 371–378 (2014)
References 187
85. Zhang, B.-L., Feng, A.-M., Li, J.: Observer-based optimal fault-tolerant control for offshore
platforms. Comput. Electr. Eng. 40(7), 2204–2215 (2014)
86. Li, H., Hu, S.-L.J.: Optimal active control of wave-induced vibration for offshore platform.
China Ocean Eng. 15(1), 1–14 (2001)
87. Ji, C., Li, H., Wang, S.: Optimal vibration control strategy for offshore platforms. In:
Proceedings of the International Offshore and Polar Engineering Conference, Kitakyushu,
Japan, pp. 91–96 (2002)
88. Yang, J.S.: Robust mixed H2 /H∞ active control for offshore steel jacket platform. Nonlinear
Dyn. 78(2), 1503–1514 (2014)
89. Zhang, B.-L., Tang, G.-Y.: Active vibration H∞ control of offshore steel jacket platforms
using delayed feedback. J. Sound Vib. 332(22), 5662–5677 (2013)
90. Zhang, B.-L., Ma, L., Han, Q.-L.: Sliding mode H∞ control for offshore steel jacket platforms
subject to nonlinear self-excited wave force and external disturbance. Nonlinear Anal. Real
World Appl. 14(1), 163–178 (2013)
91. Zhang, B.-L., Huang, Z.-W., Han, Q.-L.: Delayed non-fragile H∞ control for offshore steel
jacket platforms. J. Vib. Control 21(5), 959–974 (2015)
92. Zhou, Y.-J., Zhao, D.-Y.: Neural network-based active control for offshore platforms. China
Ocean Eng. 17(3), 461–468 (2003)
93. Chang, S., Kim, D., Chang, C., et al.: Active response control of an offshore structure under
wave loads using a modified probabilistic neural network. J. Mar. Sci. Technol. 14(2), 240–
247 (2009)
94. Kim, D.H.: Neuro-control of fixed offshore structures under earthquake. Eng. Struct. 31(2),
517–522 (2009)
95. Kim, D.H.: Application of lattice probabilistic neural network for active response control of
offshore structures. Struct. Eng. Mech. 31(2), 153–162 (2009)
96. Cui, H., Hong, M.: Adaptive inverse control of offshore jacket platform based on grey
prediction. In: Proceedings of the International Conference on Digital Manufacturing and
Automation, Zhangjiajie, China, pp. 150–154 (2011)
97. Li, X., Yu, X., Han, Q.-L.: Stability analysis of second-order sliding mode control systems
with input-delay using Poincare map. IEEE Trans. Autom. Control 58(9), 2410–2415 (2013)
98. Zhang, B.-L., Tang, G.-Y., Ma, H.: Optimal sliding mode control with specified decay rate for
offshore steel jacket platforms. China Ocean Eng. 24(3), 443–452 (2010)
99. Zhang, B.-L., Han, Q.-L., Zhang, X.-M., et al.: Integral sliding mode control for offshore steel
jacket platforms. J. Sound Vib. 331(14), 3271–3285 (2012)
100. Zhang, X.-M., Han, Q.-L., Han, D.-S.: Effects of small time-delays on dynamic output
feedback control of offshore steel jacket structures. J. Sound Vib. 330(16), 3883–3900 (2011)
101. Zhang, B.-L., Han, Q.-L., Zhang, X.-M., et al.: Sliding mode control with mixed current and
delayed states for offshore steel jacket platforms. IEEE Trans. Contr. Syst. Technol. 22(5),
1769–1783 (2014)
102. Nourisola and Ahmadi Nourisola, H., Ahmadi, B.: Robust adaptive sliding mode con-
trol based on wavelet kernel principal component for offshore steel jacket platforms
subject to nonlinear wave-induced force. J. Vib. Control (2014). https://doi.org/10.1177/
1077546314553319
103. Nourisola, H., Ahmadi, B., Tavakoli, S.: Delayed adaptive output feedback sliding mode
control for offshore platforms subject to nonlinear wave-induced force. Ocean Eng. 104, 1–9
(2015)
104. Robinett, R.D., Petterson, B.J., Fahrenholtz, J.C.: Lag-stabilized force feedback damping. J.
Intell. Robot. Syst. 21(3), 277–285 (1998)
105. Zhao, Y.-Y., Xu, J.: Effects of delayed feedback control on nonlinear vibration absorber
system. J. Sound Vib. 308(1–2), 212–230 (2007)
106. Zhang, D., Han, Q.-L., Jia, X.-C.: Network-based output tracking control for a class of T-S
fuzzy systems that can not be stabilized by non-delayed output feedback controllers. IEEE
Trans. Cybern. 45(8), 1511–1524 (2015)
188 References
107. Zhang, B.-L., Han, Q.-L.: Robust sliding mode H∞ control using time-varying delayed states
for offshore steel jacket platforms. In: Proceedings of the IEEE International Symposium on
Industrial Electronics, Taipei, Taiwan, pp. 1–6 (2013)
108. Zhang, B.-L., Han, Q.-L., Huang, Z.-W.: Pure delayed non-fragile control for offshore steel
jacket platforms subject to non-linear self-excited wave force. Nonlinear Dyn. 77(3), 491–502
(2014)
109. Sakthivel, R., Selvaraj, P., Mathiyalagan, K., et al.: Robust fault-tolerant H∞ control for
offshore steel jacket platforms via sampled-data approach. J. Franklin Ins. 352(6), 2259–2279
(2015)
110. Sakthivel, R., Santra, S., Mathiyalagan, K., et al.: Robust reliable sampled-data control for
offshore steel jacket platforms with nonlinear perturbations. Nonlinear Dyn. 78(2), 1109–
1123 (2014)
111. Sivaranjani, K., Rakkiyappan, R., Lakshmanan, S., et al.: Robust stochastic sampled-data
control for offshore steel jacket platforms with non-linear perturbations. IMA J. Math. Control
Info. (2015). https://doi.org/10.1093/imamci/dnv046
112. Zhang, B.-L., Meng, M.-M., Han, Q.-L., et al.: Robust non-fragile sampled-data control for
offshore steel jacket platforms. Nonlinear Dyn. 83(4), 1939–1954 (2016)
113. Huang, S., Cai, M., Xiang, Z.: Robust sampled-data H∞ control for offshore platforms subject
to irregular wave forces and actuator saturation. Nonlinear Dyn. (2017). https://doi.org/10.
1007/s11071-017-3404-6
114. Peng, C., Han, Q.-L., Yue, D.: To transmit or not to transmit: a discrete event-triggered
communication scheme for networked Takagi-Sugeno fuzzy systems. IEEE Trans. Fuzzy
Syst. 21(1), 164–170 (2013)
115. Zhang, X.-M., Han, Q.-L.: Event-triggered dynamic output feedback control for networked
control systems. IET Control Theory & Appl. 8, 226–234 (2014)
116. Zhang, X.-M., Han, Q.-L.: Event-based H∞ filtering for sampled-data systems. Automatica
51, 55–69 (2015)
117. Ge, X., Yang, F., Han, Q.-L.: Distributed networked control systems: a brief overview. Inf.
Sci. 380, 117–131 (2017)
118. Zhang, X.-M., Han, Q.-L., Yu, X.: Survey on recent advances in networked control systems.
IEEE Trans. Ind. Informat. 12(5), 1740–1752 (2016)
119. Zhang, X.-M., Han, Q.-L., Zhang, B.-L.: An overview and deep investigation on sampled-
data-based event-triggered control and filtering for networked systems. IEEE Trans. Ind.
Informat. 13(1), 4–16 (2017)
120. Zhang, B.-L., Han, Q.-L.: Network-based modelling and active control for offshore steel
jacket platforms with TMD mechanisms. J. Sound Vib. 333(25), 6796–6814 (2014)
121. Jiang, X., Han, Q.-L.: On H∞ control for linear systems with interval time-varying delay.
Automatica 41(12), 2099–2106 (2005)
122. Jiang, X., Han, Q.-L.: Delay-dependent robust stability for uncertain linear systems with
interval time-varying delay. Automatica 42(6), 1059–1065 (2006)
123. Zhang, B.-L., Han, Q.-L., Zhang, X.-M.: Event-triggered H∞ reliable control for offshore
structures in network environments. J. Sound Vib. 368, 1–21 (2016)
124. Yue, D., Tian, E., Han, Q.-L.: A delay system method for designing event-triggered controllers
of networked control systems. IEEE Trans. Automa. Control 58(2), 475–481 (2013)
125. Peng, C., Han, Q.-L.: A novel event-triggered transmission scheme and L2 control co-design
for sampled-data control systems. IEEE Trans. Automa. Control 58(10), 2620–2626 (2013)
126. Sarpkaya, T., Isaacson, M. (eds.): Mechanics of Wave Forces on Offshore Structures. Van
Nostrand Reihhold, New York (1981)
127. Chakrabarti, S.K. (ed.): Hydrodynamics of Offshore Structures. Springer, Berlin (1987)
128. Xie, L.: Output feedback H∞ control of systems with parameter uncertainty. Int. J. Control
63(4), 741–750 (1996)
129. Lancaster, P., Lerer, L., Tismenetsky, M.: Factored forms for solutions of AX − XB = C and
X − AXB = C in companion matrices. Linear Algebra Appl. 62, 19–49 (1984)
References 189
130. Gahinet, P., Apkarian, P.: A linear matrix inequality approach to H∞ control. Int. J. Robust
Nonlinear Control 4, 421–448 (1994)
131. Han, Q.-L.: Absolute stability of time-delay systems with sector-bounded nonlinearity.
Automatica 41, 2171–2176 (2005)
132. Zhang, X.M., Wu, M., She, J.H., et al.: Delay-dependent stabilization of linear systems with
time-varying state and input delays. Automatic 41(8), 1405–1412 (2005)
133. Peng, C., Fei, M.-R.: An improved result on the stability of uncertain T-S fuzzy systems with
interval time-varing delay. Fuzzy Sets Syst. 212, 97–109 (2012)
134. Zhang, X.M., Han, Q.-L.: Novel delay-derivative-dependent stability criteria using new
bounding techniques. Int. J. Robust Nonlinear control 23(13), 1419–1432 (2013)
135. Spurgeon, S., Edwards, C. (eds.): Sliding Mode Control: Theory and Applications. Taylor and
Francis, London (1998)
136. Yu, X., Kaynak, O.: Sliding mode control with soft computing: a survey. IEEE Trans. Ind.
Electron. 56(9), 3275–3285 (2009)
137. Ghaoui, L.E., Oustry, F., AitRami, M.: A cone complementarity linearization algorithms for
static output feedback and related problems. IEEE Trans. Automa. Control 42(8), 1171–1176
(1997)
138. Robinett, R.D., Petterson, B.J., Fahrenholtz, J.C.: Lag-stabilized force feedback damping. J.
Intell. Robot. Syst. 21(3), 277–285 (1998)
Index
A Fixed offshore platform, 5, 10

Active control, 2, 6–14, 17, 18, 22, 23, 47, 48, Floating platform, 6
71, 89, 108, 128, 129, 131, 132, 153, Floating structure, 1, 10
156, 181 Friction damper, 2, 3, 5
Active mass damper (AMD), 8, 9, 11, 12, 14, Fuzzy logic control, 7
15, 17–21, 33, 38, 68, 107, 155, 156,
167, 169
Actuator, 7, 9, 13–15, 115, 132, 133, 155–157, H
166, 169, 170, 174, 176–181 H∞ control, 9–15, 49–69, 89, 91–108, 155,
AMD, see Active mass damper 166, 169–180
Asymptotically stable, 21, 34, 51, 53, 54, 76, H2 control, 9, 10
98, 99, 110, 135, 138, 152, 158, 159,
161–163, 166–168
I
C
Ice-induced vibration, 3–5
Cone complementary linearization (CCL)
Integral sliding mode control (ISMC), 11, 14,
algorithm, 72
15, 49–69, 71–89, 91, 99, 106, 107,
Controller, 7, 9–15, 21, 33–37, 39–49, 53, 58,
128, 131, 151, 153
62, 67–69, 71–77, 80, 82–89, 101–103,
Intelligent control, 2, 10, 69
105–107, 109–113, 115–129, 131–153,
155–169, 171, 173, 174, 177, 179–181
Controller design, 2, 7–15, 22, 48, 49, 88,
116–121, 131, 134–140, 180 J
Jacket platform, 1, 3–13, 15, 17, 22, 27, 33,
D 34, 37, 42, 47–49, 68, 71, 89, 91, 100,
Delayed feedback control, 2, 12–13, 89, 108, 107–109, 131, 132, 134, 180, 181
128, 129
Disorder packet, 153
L
E Linear matrix inequalities (LMIs), 77, 89, 96,
Event-triggered communication scheme, 159 97, 111, 113, 119, 121, 122, 168
Liquid column dampers, 6
F LMI, see Linear matrix inequalities
Fault-tolerant control, 9, 181 Lyapunov Krasovskii functional candidate, 14,
Feedforward control, 33–48 51, 53, 73, 93, 117, 135, 136, 160

https://doi.org/10.1007/978-981-13-2986-9
192 Index
N Seismic response, 2
Networked control systems (NCSs), 152 Semi-active control, 1, 6–7, 181
Networked-induced delays, 14, 133, 140–145, Sensor, 7, 13, 115, 132, 133, 181
150–153, 156, 170, 174, 176, 177, 180, Sliding mode control (SMC), 2, 10–11, 15, 49,
181 54–56, 58, 59, 62, 63, 67, 68, 71, 76,
Neural network, 7, 10, 68 79–82, 87–89, 107, 128, 146, 148–150
Non-fragile control, 13, 92, 93, 101, 107 Stability, 2, 7, 8, 10, 14, 51–53, 74, 94, 112,
119, 131, 134, 136, 155, 159, 164, 167,
180, 181
O Stability analysis, 49–52, 134–140, 153
Ocean engineering, 181 Stabilization, 107, 177, 179
Offshore platform, 1–15, 17–31, 33, 36, 38–41, Stochastic control, 48
43–49, 56–58, 63, 67–69, 71, 78, 79, 82,
83, 86–89, 91, 92, 100, 103, 107–110,
116, 123, 126–129, 131, 132, 134, 138,
T
140–142, 144–148, 150–153, 156, 169,
Tension leg platform (TLP), 1, 4–6, 8, 10
171, 172, 179–181
Time-delay, 2, 9, 12, 13, 15, 71, 79–83, 86–89,
Offshore structure, 1, 18–20, 132, 153, 155,
91, 92, 102–104, 106–109, 115, 116,
156, 167, 169–171, 173, 174, 176, 177,
124, 126–128, 133, 157
179, 180
TLP, see Tension leg platform
Optimal control, 2, 7–10, 33, 36–38, 44–48,
Tracking control, 9, 15, 21, 31–48, 173
68, 173
Tuned mass damper (TMD), 4–5, 8, 9, 11–15,
Output feedback control, 9, 11–13, 91, 109
17, 22–29, 56, 57, 78, 107, 109, 113,
131, 132, 140, 153
P
Passive control, 1–7
V
Vibration control, 1, 2, 5–7, 10, 15, 181
R Vibration mitigation, 5
Robust controller design, 89 Vibration reduction, 4, 11
Robust H∞ control, 10, 49, 54–56, 62, 89,
91–108
Robust stability analysis, 94
W
Wave load, 5
S
Sampled-data control, 2, 13, 129
Schur complements, 29, 51, 74, 95, 113, 120, Z
138, 163, 164 Zero order hold (ZOH), 156, 176

Ships

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Ships

Transféré par

Droits d'auteur :

Formats disponibles

Bao-Lin Zhang · Qing-Long Han

Active Control of Offshore

Xian-Ming Zhang Gong-You Tang

ISBN 978-981-13-2985-2 ISBN 978-981-13-2986-9 (eBook)

Library of Congress Control Number: 2018961387

© Springer Nature Singapore Pte Ltd. 2019

Optimal tracking control Chapter 3 presents an optimal tracking control scheme

Hangzhou, China Bao-Lin Zhang

4 Integral Sliding Mode H∞ Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

8 Network-Based Modeling and Active Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

© Springer Nature Singapore Pte Ltd. 2019 1

1.1 Passive Control

1.1.1 Hysteretic and/or Viscoelastic Mechanisms

specified that as an intelligent material, the shape memory alloy is of advantages

1.1.2 Damping Isolation Mechanisms

To dissipate unwanted vibration of offshore platforms, designing an isolation

It is demonstrated from experimental and computational results that the proposed

1.1.3 Dynamic Vibration Absorbers

1.1.3.1 TMD Mechanisms

A TMD device is composed of a mass, a spring, and a viscous damper mechanism.

1.1.3.2 TLD Mechanisms

Due to effectiveness of vibration mitigation, low cost of manufacturing, convenience

and closed on top of a dual chamber. It is shown theoretically and experimentally

1.2 Semi-active Control

Note that semi-active control is a characteristic of controllable damping over contin-

control is widely used in engineering vibration areas. Specifically, it requires less

1.3 Active Control

In practical applications, a passive control device usually exhibits its limitations,

1.3.1 Optimal Control

As an advanced control technology, optimal control aims at optimizing a certain

natural frequency of offshore platforms [11]. Based on a single-degree-of-freedom

1.3.2 Robust Control

Robust control is to design proper state/output feedback controllers against uncer-

and external disturbance. Recently, by considering controller perturbations caused

1.3.3 Sliding Mode Control

and earthquakes. As an efficient method to cope with parameter uncertainties and

1.3.4 Delayed Feedback Control

Under some circumstances, by properly introducing time-delays into control chan-

1.3.5 Network-Based Control

Network-based control provides an advanced remote control strategy, where control

1.4 Book Outline

The book is organized as follows: Chap. 1 presents an overview of recent devel-

2.1 Model of an Offshore Platform with AMD Mechanisms

Consider a simplified offshore steel jacket platform with an AMD mechanism

© Springer Nature Singapore Pte Ltd. 2019 17

Fig. 2.1 An offshore steel jacket structure with an AMD [79]

x(t) = [x1 (t) x2 (t) x3 (t) x4 (t)]T

Then, the motion equation (2.1) can be written as

ẋ(t) = Ax(t) + Bu(t) + Df (t), x(0) = x0 (2.3)

where x0 is the initial state vector, and

where Hs denotes the significant wave height, γ̄ ∈ [0, Hs ] is the peakedness

with θ = 0.07 for ω ≤ ω0 and θ = 0.09 for ω > ω0 .

υ(z, t) = Tυη (ω, z)η(t) (2.7)

where ρ is the fluid density, D̃ represents the cylinder diameter, Cd is drag

where φ(z) is the shape function as

ηj (t) = Aj cos(−ωj t + ςj ) (2.15)

ẇ(t) = Gw(t), f (t) = H w(t) (2.18)

Re(λ) = 0, ∀λ ∈ σ (G) (2.20)

2.2 Model of an Offshore Platform with Active TMD

⎪ − φ2 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + φ2 [ζ2 (t) − u(t)]

Notice that K̂T KT leads to

ΔωT (t) = ωT ω̂T · Δω̃T (t) + o(Δω̃T (t)) (2.25)

Δωi (t) = ωi ω̂i · Δω̃i (t) + o(Δω̃i (t)), i = 1, 2 (2.27)

f (x, t) ≤ μ x (2.43)