0 Votes +0 Votes -

2 vues12 pagesJan 29, 2018

© © All Rights Reserved

PDF, TXT ou lisez en ligne sur Scribd

© All Rights Reserved

2 vues

© All Rights Reserved

- EE4CL4_lecture30
- Control Final Exam Model. B..
- PID Tuning Tutor
- Topic1 Control System
- State Feedback Control of Inverted Pendulum
- Main Steam Turbine Controls Retrofit ISA Final
- re30131_0203
- OGATA_doc
- Control and Simulation in LabVIEW 3-58 SMCIE TACN 2013
- Dynamic Control of Electric Power Systems
- jgcd
- Dynamic Control of Electric Power Systems
- Solns Adaptive Control by Astrom
- Adaptive Control Design for High-Order MIMO Nonlinear Time-Delay Systems Based on Neural Network
- 2013_2014_report1_copy (1)
- 0071462775_ar091.PDF Control Systems Based in Ogata Book
- Introduction to Controls
- Decentralized Design of Feedback Control of Large Scale System
- Pid
- Control System Introduction

Vous êtes sur la page 1sur 12

www.elsevier.com/locate/arcontrol

Lubomı́r Bakule

Institute of Information Theory and Automation, Academy of Sciences of the Czech Republic, 182 08 Prague 8, Czech Republic

Received 8 October 2007; accepted 4 March 2008

Available online 2 May 2008

Abstract

The paper reviews the past and present results in the area of decentralized control of large-scale complex systems. An emphasis is laid on

decentralization, decomposition, and robustness. These methodologies serve as effective tools to overcome specific difficulties arising in large-

scale complex systems such as high dimensionality, information structure constraints, uncertainty, and delays. Several prospective topics for future

research are introduced in this contents. The overview is focused on recent decomposition approaches in interconnected dynamic systems due to

their potential in providing the extension of decentralized control into networked control systems.

# 2008 Elsevier Ltd. All rights reserved.

1. Introduction Uncertainty;

Delays.

The notion of large-scale systems has been introduced when it

became clear that there are real world control problems that The theory of large-scale systems is devoted to the problems

cannot be solved by using one-shot approaches. Such typical that arise from above difficulties. The theory answers the funda-

motivating problems arise in the control of interconnected power mental questions of how to break down a given control problem

systems with strong interactions, water systems which are widely into manageable subproblems which are only weakly related to

distributed in space, traffic systems with many external signal, or each other and can be solved independently. These general facts

large-space flexible structures. The reason is that the systems to lead to the opinion that no formal definition of the term large-scale

be controlled are too large and the problems to be solved are too system is necessary. Instead, a more pragmatic view has been

complex. That is, these tasks cannot be solved simply by using adopted: A system is considered large-scale if it is necessary to

faster computers with larger memory. They necessitate new ideas partition the given analysis or synthesis into manageable sub-

for dividing the analysis and synthesis of the overall system into problems. As a result, the overall plant is no longer controlled by a

independent or almost independent subproblems, for dealing single controller but by several independent controllers which all

with the incomplete information about the system, for treating together represent a decentralized controller. This is the funda-

with the uncertainties, and for dealing with delays. The mental difference between feedback control of small and large

complexity is an essential and dominating problem in systems systems usually described by the idea of information structure.

theory and practise. It leads to severe difficulties that are The basic problem is to find control inputs uðtÞ on the basis

encountered in the tasks of analyzing, designing, and imple- of the a priori knowledge of the plant S described by its model

menting appropriate control strategies and algorithms. These M for a supposed class of disturbances wðtÞ and the control goal

difficulties arise mainly from the following well-known reasons: given in the form of the design requirements DM, and the a

posteriori information about the outputs yðtÞ and the command

Dimensionality; signals rðtÞ. The design problem is completely described by

Information structure constraints; the information given by these four quantities. The problem

consists of two phases:

§

Preliminary version of this paper was presented at the 11th IFAC/IFORS/

IMACS/IFIP Conference on Large-Scale Systems: Theory and Applications, Design phase: Determine control laws on the basis of the a

2007, Gdansk, Poland. priori information about the plant and the design requirements.

1367-5788/$ – see front matter # 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.arcontrol.2008.03.004

88 L. Bakule / Annual Reviews in Control 32 (2008) 87–98

posteriori information about the system state delivered by the

outputs and the given command signals.

control as illustrates Fig. 1 (a), while non-classical information

structure of decentralized control is shown in Fig. 1(b). The

decision makers and the controllers have available only parts of

the overall a priori and a posteriori information in decentralized

control (Bakule & Lunze, 1988; Lunze, 1992).

large-scale systems distinguished by the degree to which they

reflect the internal structure of the overall dynamic system.

These structures are called multi-channel systems and

interconnected systems as illustrated in Fig. 2 (b) and (c).

Multi-channel systems input and output vectors are decom- Fig. 2. System structures: (a) centralized system; (b) N-channel system; (c)

posed into subvectors, while the system is considered as one interconnected system.

whole. Interconnected systems operate with interactions

between subsystems. They are represented by signals through which subsystems interact among themselves. These signal are

internal signals of the overall system.

To cope with the aforementioned appearance of the

complexity issues several general methodologies have been

and are being elaborated. Most of them belong to one of the

following three groups (Šiljak, 1978):

Decentralization;

Decomposition;

Robustness and model simplification.

in the solution of the given decision problem. The decentraliza-

tion of the control law concerns on-line information about the

state and the command. It enables a completely independent

implementation of the control stations. The decentralization of

the process refers to the model and the design goals. It supports a

way in which the control stations are found independently by

solutions of separate design process. There is a variety of

different motivating reasons for the decentralization of the design

process such as for instance weak coupling of subsystems,

subsystems have contradictory goals, subsystems are assigned to

different authorities, or the high dimensionality of the overall

system. The principal ways of decentralizing the design tasks

belong to two groups (Lunze, 1992):

Decentralized design for weakly coupled subsystems.

means that at least an approximate model of all other subsystems

must be considered for the design of any subsystem under the

current design, while the coupling can be neglected during

the design of individual control stations when considering the

Fig. 1. Control design: (a) classical information structure; (b) non-classical decentralized design for weakly coupled subsystems. This

information structure. situation is illustrated in Fig. 3.

L. Bakule / Annual Reviews in Control 32 (2008) 87–98 89

involving linear matrix inequalities (LMIs) (Boyd, El Ghaoui,

Feron, & Balakrishnan, 1994). Parametric uncertainties are

supposed within both interconnected systems and the designed

controllers. Thus, the resulting LMIs lead to the design of

resilient decentralized controllers with additive uncertainties.

The structures of gain matrices in decentralized control

correspond with some well-known forms introduced in the

theory of sparse matrices (Tewarson, 1973). A block-diagonal

form (BD) is used when considering disjoint subsystems or

symmetric composite systems, while a block-tridiagonal form

(BTD) corresponds with overlapping subsystems. A bordered

block-diagonal form (BBD) is desirable in distributed control

and in applications with communication overhead (Zečević &

Šiljak, 2005a). The survey considers three prototype problems:

The decentralized stabilization of nonlinear disjoint subsys-

tems, H 1 control design of resilient output controllers for

Fig. 3. Decentralization of the design task for subsystem 1: (a) weak interac- overlapping subsystems, and H 2 control design of delayed

tions; (b) strong interactions. symmetric composite systems. An extension to decentralized

control design for interconnected subsystems with a commu-

nication overhead is included. References for delay free

Several decomposition approaches have been elaborated to systems and delayed systems are distinguished.

simplify the analysis and synthesis tasks for large-scale The control stations can be designed in connection with the

systems. The main goal of the decomposition is the reduction subsystem model only in weakly coupled subsystems. It means

of computational complexity. These approaches can classified that the design problems are completely independent for all

as follows: controllers. The controllers will behave similarly if the

interconnections are sufficiently weak. Conceptual insight

Disjoint subsystems; and useful solutions of problems based on this rather primitive

Overlapping subsystems; division of the global design problem are almost exhausted.

Symmetric composite systems; Recently, Becerril, Aghdam, and Yurkevich (2007) have

Multi-time scale systems; applied generalized sampling on decentralized control of

Hierarchically structured systems. two-time scale systems and Aghdam and Davison (2007) have

proposed decentralized switching control for hierarchical

The first three items belong to strongly coupled subsystems, systems. Cantoni et al. (2007) proposed decentralized as well

while the last two items are weakly coupled subsystems as distributed control for irrigation networks modeled as a

(Bakule & Lunze, 1988; Lunze, 1992; Šiljak, 1991). hierarchical system. The use the standard sequential design,

Robustness analysis has to exploit the character of where the output from the previous pool is considered as a

uncertainties mainly on the bases of the stability analysis of known disturbance entering into the current pool under control

coupled systems, while model simplification includes mainly design. The main future trend in decentralized control should

model reduction methods and approximations (Šiljak & concern more sophisticated and advanced division approaches

Zečević, 2005). which correspond with strongly coupled systems. It motivates

Delays are much more important than the accuracy of the to restrict this survey on the classes of strongly coupled

transmitted information due to the fact that feedback control subsystems.

systems are quite robust to such inaccuracy. The presence of The origin and the rapid development of decentralized

delays is unavoidable under current technology trends of shared control design methods began since the 1970s. Various

digital networks and wireless connections. A key role plays the decentralized control design structures and algorithms have

structure and implementation issues of feedback control. been developed to present the flexibility and superiority of this

Decentralized control has the potential of being superior over approach for different classes of interconnected systems.

centralized control also in this respect, because sharing local The presentation cannot be encyclopedic due to space

information may be relatively delay free (Baillieul & Antsaklis, limitation. A large number of valuable and promising results

2007; Zhang & Xie, 2007). had to be omitted though they deserve to be included in any

overview presenting decentralized control issues. There is

1.2. Outline of the paper available a variety of monographs and survey papers including

numerous references therein (Bakule & Lunze, 1988;

The paper is focused on strongly coupled subsystems. It Bernussou & Sebe, 2002; Gajić & Ikeda, 2004; Gajić & Shen,

surveys the recent achievements in decentralized control 1993; Jamshidi, 1997; Jiang, 2003; Lunze, 1992; Šiljak, 1991,

design emphasizing nonlinearities, uncertainties, and delays 1996; Šiljak & Zečević, 1999, 2005; Tamura & Yoshikawa,

90 L. Bakule / Annual Reviews in Control 32 (2008) 87–98

1990), which offer a more wider and deeper view to the readers DAdi j ðtÞ are norm bounded uncertainties which admit the

interested in decentralized control theory and practice. standard structure:

Note only that the majority of references consider zero

DAðÞ ðtÞ ¼ DðÞ F ðÞ ðtÞEðÞ (3)

reference signals corresponding with a standard requirement on

the closed-loop system stability under uncertainties and delays. where ðÞ denotes the corresponding subindices in all DA in (1)

The decentralized servomechanism problem for nominally and (2). DðÞ and EðÞ are constant matrices corresponding with

linear interconnected systems with structured parameter the matrix DAðÞ ðtÞ. Uncertainties are lumped in unknown

perturbations solved Vaz and Davison (1989) for non-zero Lebesgue measurable functions F ðÞ satisfying the bounds.

reference signals. Global decentralized output regulation for a T

FðÞ F ðÞ I for all t 0. I denotes a unit matrix of appropriate

class of uncertain interconnected systems with nonlinear dimension.

exosystem derived for the first time Xi and Ding (2007). The goal is to find a decentralized resilient controller-

Cantoni et al. (2007) use local non-zero reference signals observer in the form:

related to individual locally controlled pools of an open-water

channel. x̂˙ i ðtÞ ¼ Ai x̂i ðtÞ þ Bi ui ðtÞ þ ½K oi þ DK oi ½yi ðtÞ C i x̂i ðtÞ;

(4)

ui ðtÞ ¼ ½K ci þ DK ci x̂i ðtÞ; i ¼ 1; . . . ; N

2. Disjoint subsystems

where x̂i ðtÞ is the ni -dimensional controller state of the

The actual tearing of the system may be performed from subsystem i. DK oi ¼ Doi F oi ðtÞEoi , DK ci ¼ Dci F ci ðtÞEci are

either conceptual or numerical reasons. Conceptual reasons the i th observer and controller gain perturbations.

correspond usually with the boundaries of physical subsystems. Doi ; Eoi ; Dci ; Eci are given constant matrices. F ðÞ ðtÞ are

Numerical reasons require to develop a universal decomposi- unknown arbitrarily time-varying Lebesgue measurable func-

tion technique. It leads to the notion of e decomposition when tions satisfying the relation F ðÞ ðtÞT F ðÞ ðtÞ I. K oi ; K ci are

considering disjoint subsystems. The idea od epsilon decom- the controller-observer gain matrices to be determined for all

position can be simply explained on a linear dynamic system i ¼ 1; . . . ; N.

T

ẋ ¼ ðAD þ eAC Þx, where the matrix AD ¼ diagðA1 ; . . . ; AN Þ, Consider xe ¼ ½xT ; eT , where eðtÞ ¼ xðtÞ x̂ðtÞ denotes

the matrix AC has all its elements smaller than one, and e is a the error vector. x and x̂ are the overall system state vector and

prescribed small number. The choice of e influences on the the overall controller state vector. The overall augmented

strength of interconnections. If each subsystem Ai is stable, then systems (1)–(4) result in

an appropriate choice of e preserves the weak coupling property

x̄e ðtÞ ¼ ½Ā þ DĀxe ðtÞ þ ½Ād þ DĀd xe ðt dÞ (5)

of the system and thereby the stability of the overall system.

The increasing threshold of e leads to the notion of nested e where

decomposition (Šiljak, 1991, 1996).

Consider the stabilization problem for N interconnected A þ BK c BK c Ad 0

Ā ¼ ; Ād ¼ ;

subsystems with parameter uncertainties and delays with a 0 A KoC Ad 0

given decomposition in the form: DA þ BDK c BDK c C DAd 0

DĀ ¼ ; DĀd ¼

DA DK o C DAd 0

S : ẋi ðtÞ

(6)

¼ ½Ai þ DAi ðtÞxi ðtÞ þ ½Adi þ DAdi ðtÞxi ðt dÞ þ Bi ui ðtÞ

Note that the matrices K o and K c in (6) are block diagonal

þ si ðtÞ; matrices. The stability of the system (5)–(6) is established

by using the Lyapunov–Krasovski functional when consider-

yi ðtÞ ¼ C i xi ðtÞ; xi ðto Þ ¼ Fi ðto Þ; d to 0; ing the Lyapunov functionals for individual subsystems.

These individual functions result in a global Lyapunov

i ¼ 1; . . . ; N

functional

(1) Z t

where xi ðtÞ; ui ðtÞ; si ðtÞ; yi ðtÞ are ni -, mi -, pi -, r i -dimensional Vðxe ; tÞ ¼ xe ðtÞT Pe x̄m ðtÞ þ xe ðsÞT Qe xe ðsÞ ds (7)

td

vectors of the subsystem states, control inputs, interconnection

inputs and measured outputs, respectively. Fi ðto Þ is a given where Pe 2 R2n2n > 0 and Qe 2 R2n2n P> 0 with Pe ¼

initial function. diagðPs ; Pc Þ and Qm ¼ diagðQs ; Qc Þ, n ¼ i ni . The matrices

Interconnections are described in the form: Ps ; Pc ; Qs ; Qc have a block diagonal structure corresponding

with individual subsystems.

X

N

Taking the total time derivative of V with respect to the

si ðtÞ ¼ ½Ai j þ DAi j ðtÞx j ðtÞ þ ½Adi j þ DAdi j ðtÞx j ðt dÞ

j¼1

system (5) and (6) leads, after lengthy but straightforward

computations, to a sufficient condition given in the form the

(2)

LMIs for given Pe and Qe . Suppose there exist block diagonal

where d denotes a point time delay. Ai ; Adi ; Bi ; C i ; Ai j , and Adi j matrices Y s ; Y c ; Y h and positive scalars e1 ; . . . ; e4 such the

are constant nominal matrices. DAi ðtÞ, DAdi ðtÞ, DAi j ðtÞ, and LMIs

L. Bakule / Annual Reviews in Control 32 (2008) 87–98 91

2 3 2 3

Ws Pa Ps X s Qs Ad Wc Pc ZcT X s Qs Ad 2 3

6 Wd 2DA 2DAd 2Ad

6 Ls 0 0 0 77

6

6 Lc 0 0 0 77 6

6 2e1 I 0 0 7

6 Ls 0 0 77 < 0;

6

6 e3 I 0 0 77 < 0;

6

4

7<0 (8)

4 2e4 I 0 5

Qs 0 5 4 Qs 0 5

Sd

Ss Ss

have a feasible solution. The block matrices used in (8) mean robust control problem are solved in Chen (1989, 1992) and Chen

W s ¼ AX s þ X s AT þ BY s þYsT B, W c ¼ AX c þ X c AT þ Y c þ YcT , and Han (1993). Decentralized adaptive control provide Shi and

W d ¼ BY d þ YdT BT þ Y h þ YhT , Ls ¼ ðe1 I; e2 IÞ, P a ¼ ðDA ; Dc Þ, Singh (1992) for systems with strong nonlinear interconnections

P s ¼ ðZaT ; ZbT Þ, Lc ¼ ðe1 I; e3 IÞ, P c ¼ ðDA ; Do ; DAd Þ, Ss ¼ and Wu (2003) for uncertain interconnections. Decentralized

T T

Qs þ e4 EAd EAd , and Sd ¼ 2Qs e4 EAd EAd . The necessary adaptive stabilization for interconnected stochastic systems

linearization matrices used in the derivation of (8) are proposed Liu, Zhang and Jiang (2007) using the concept of input-

X s ¼ P1 1 1 T 1

s , X c ¼ Pc , Y h ¼ e2 Ps Ec Ec Ps , Z a ¼ e1 EA Ps ,

1

to-state stability. An another interesting and practically important

1 1

Z b ¼ e2 Ec Pc , Z c ¼ e3 Ec CPc . result is decentralized adaptive stabilization of unknown

The resulting block diagonal gain matrices stabilizing the interconnected systems with hysteresis solve using the back-

system (1)–(4) have the following form: stepping technique by Wen and Zhou (2007). Decentralized

stabilization is applied to flexible structures in Li, Kosmato-

1

K c ¼ Y s Xs1 ; K o ¼ Y c Xc1 CT ðCC T Þ (9) poulos, Ioannou, Ryaciotaki-Boussalis (2000) and Kobayashi,

Ikeda, and Fujisaki (2007) and to power systems Befekadu and

The above prototype design of decentralized robust Erlich (2006). Robust decentralized control independent design

resilient stabilizing controller is developed for the systems developed Kozáková and Veselý (2006), while Rosinová and

with state delays as well as norm bounded uncertainties in Veselý (2006) proposed the PID design by using LMIs.

both the system and the observer-based controller gain Extensions to descriptor systems are presented in Wang and

matrices. The motivating reasons why to include controller Bao (2000) and Chen, Gui, and Zhai (2008). New results have

uncertainties into the controller design are explained in been obtained in decentralized output feedback control by

Istepanian and Whidborne (2001) and Mahmoud (2004). The Orqueda, Zhang and Fierro (2007), Lee (2007), Xi and Ding

importance of non-fragile controllers with information (2007), and Polendo and Qian (2007). Another interesting results

structure constraints is underlined when considering large- consider Borrelli et al. (2005) for hybrid systems, while Richards

scale systems. and Chow (2007) present predictive control approach for systems

Delay free systems include a number of interesting with coupled constraints.

recent results concerning designs of decentralized control of Delayed systems with state delays consider Mukaidani,

interconnected systems with disjoint subsystems which are Tanaka and Xu (2003) by using LMIs to solve a guaranteed cost

formulated as standard convex programming problems. This control problem, while Kown and Park (2006) solve the same

approach offers a possibility to apply a variety of algorithms problem by using LMIs when considering delay in the feedback

available in the linear or bilinear matrix inequalities framework. loop. Decentralized control for dynamic routing in commu-

Various recent results of this type include uncertain and nonlinear nication network proposed PIftar and Davison (2002). Kown

systems satisfying quadratic constraints. Both structural and and Park (2006) introduce delays within the framework of

algebraic enhancements of decentralized control with state or distributed control.

static output feedback are presented in Šiljak and Stipanović

(2000a, 2001), Šiljak and Zečević (2005), and Zečević and Šiljak 3. Overlapping subsystems

(2004). An extension of these results on decentralized dynamic

output feedback control design are considered in Pagilla and The decomposition of the overall system into disjoint

Zhou (2005) and Stanković, Stipanović, and Šiljak (2007). subsystems is not effective if the subsystems are strongly

Zečević and Šiljak (2005a) introduced a BBD form for gain coupled. It means that a given system has no epsilon

matrices. It corresponds in a certain manner with an extension of decomposition. Then, overlapping decomposition can be used

traditional decentralized control towards the concept of as an alternative way in which the subsystems share some

distributed control developed for spatially interconnected common parts. The overlapping subsystems may be weakly

systems in D’Andrea and Dullerud (2003), Langbort, Chandra, coupled although disjoint subsystems are not, i.e. they may

and D’Andrea (2004), and Dullerud and Andrea (2004). Chen have an overlapping epsilon decomposition (Šiljak, 1996).

et al. (2005) developed LMI type robust H 1 control design by A systematic way of overlapping decomposition means

using a homotopy method for interconnected systems, while to expand the original system with strongly coupled subsystems

Yang and Wang (1999) present the solution in terms of Riccati into a larger dimensional systems with weakly coupled

inequalities. Strong nonlinearities in both subsystems and subsystems. There is a requirement of the relation between

interactions consider within the concept of decentralized both systems. The solution of a large-dimensional system must

output-feedback control for deterministic systems Krishna- include the solution of a lower dimensional original system. A

murthy and Khorrami (2003) and Jiang (2002). Decentralized circle of ideas, methods and algorithms devoted to overlapping

92 L. Bakule / Annual Reviews in Control 32 (2008) 87–98

decompositions has been formulated rigorously into a general LMI provides a natural framework to solve the problem for

mathematical framework called the inclusion principle (Chu & state feedback (Bakule, Rodellar, & Rossell, 2005; Liu, Jiang, Su,

Šiljak, 2005; Šiljak, 1991). To recall what this principle means, & Chu, 2002) with an extension to static output feedback

let us consider the problem of H 1 control design of resilient (Zečević & Šiljak, 2004). It is important to recognize that the

output controllers for overlapping subsystems. system but with zero blocks ðÞ21 and ðÞ22 in (12) may lead to

Suppose an uncertain system with a state point delay as unfeasible LMI solution. It means that the overlapped states have

no direct access from inputs. These two generic scenarios have

S : ẋðtÞ been classified as Type I and Type II, where Type I corresponds

¼ ½A þ DAðtÞxðtÞ þ ½B þ DBðtÞuðtÞ with non zero blocks ðÞ21 and ðÞ22 while Type II considers them

as zero blocks. A suitable approach to design BTD controllers

þ ½Ad þ DAd ðtÞxðt dÞ þ B1 wðtÞ; (10) for both types of overlapping structures leads to expansion–

contraction relations for LTI delayless systems and contractibility

yðtÞ ¼ CxðtÞ; zðtÞ ¼ GxðtÞ þ HuðtÞ; xðto Þ ¼ FðtÞ; condition for controllers (Zečević & Šiljak, 2005b).

d to 0 The construction of an expanded system starts with the

transformations:

where xðtÞ 2 Rn is the state, uðtÞ 2 Rm is the control input, d > 0

is the delay time, wðtÞ 2 R p is the disturbance input and belongs x̃ðtÞ ¼ VxðtÞ; xðtÞ ¼ U x̃ðtÞ (16)

to L2 ½0; þ1Þ, zðtÞ 2 Rq is the controlled output and FðtÞ is a

where

given continuous initial function. The matrices A, B, Ad , B1 , C,

2 3

G, D are constant matrices. DðÞ ¼ DðÞ FðtÞEðÞ are real-valued I 0 0 2 3

60 I 0 0 0

norm bounded uncertainties, where DðÞ and EðÞ are given I 07

V ¼6

40

7; U ¼ 4 0 :5I :5I 05 (17)

constant matrices and FðtÞ is an unknown time-varying matrix I 05

satisfying F T ðtÞFðtÞ I. 0 0 0 I

0 0 I

Suppose a prototype case with two overlapped subsystems in

states. It means that the matrices A, DAðtÞ, Ad , DAd ðtÞ are The transformations (16) and (17) lead in a simple natural

partitioned into 3 3 blocks with the overlapping in the block way to an expanded system withTthe state vector x2 repeated in

ðÞ22 as indicate the dash lines x̃T ðtÞ= xT1 ðtÞ; xT2 ðtÞ; xT2 ðtÞ; xT3 ðtÞ . The expanded system has

the form:

¼ ½Ã þ DÃðtÞx̃ðtÞ þ ½B̃ þ DB̃ðtÞuðtÞ

The matrices B, DBðtÞ, B1 , DB1 ðtÞ possess the structure of þ ½Ãd þ DÃd ðtÞx̃ðt dÞ þ B̃1 wðtÞ; (18)

matrices partitioned into 3 2 blocks corresponding with the

ỹðtÞ ¼ C̃x̃ðtÞ; z̃ðtÞ ¼ G̃x̃ðtÞ þ HuðtÞ; x̃ðto Þ ¼ f̃ðto Þ;

overlapped states as

d to 0

matrices as

The design objective it to determine a static resilient robust C̃ ¼ CU þ Lc

BTD controller for the system (10) possessing the overlapping

structure (11), (12) and satisfying H 1 -norm bound g (13). (19)

Consider the controller K in the form: where M, N, M d , M 1 and L are complementary matrices. We

uðtÞ ¼ ½K þ DKðtÞyðtÞ (14) associate the inequality with the system S̃

(15Þ output controller

structure as K. The matrices DK ; EK are given.

L. Bakule / Annual Reviews in Control 32 (2008) 87–98 93

such that g̃ ¼ g. One possible set of conditions satisfying the and Šiljak (2003) have dealt with LTV systems. Optimization

inclusion principle has the form: issues were addressed to H 2 approaches in Ikeda et al.

(1981), Šiljak (1991), and Bakule, Rodellar, and Rossell

MV ¼ 0; M d V ¼ 0; N ¼ 0; M 1 ¼ 0; LV ¼ 0 (22) (2000a) and to H 1 approach in Bakule et al. (2005).

Recently, a variety of new results has been achieved by using

Supposing that these relation are satisfied, then the controller

the LMI approach by Šiljak and Zečević (2005), Zečević &

K̃ designed in the expanded space can be contracted into the

Šiljak (2005b) and Swarnakar, Marquez and Chen (2007).

original space as

Multi-overlapping decomposition structure consider Chen

K ¼ K̃V; DKðtÞ ¼ DK̃ðtÞV (23) and Stanković (2005). Numerous extensions of the Principle

include specialization on mechanical systems by Bakule and

The overlapping controller design is illustrated in Fig. 4. The Rodellar (1995), hybrid systems by P Iftar and Özgüner

decentralized controller design is available by using the LMI (1998), sliding mode control by Akar and Özgüner (2002),

approach in Liu et al. (2002), Bakule et al. (2005b), and Zečević and Petri nets by Aybar andP Iftar (2002). Overlapping

and Šiljak (2004). decentralized control was applied to power systems by Šiljak

Delay free systems include a large number of real world (1991), Chen and Stanković (2005, 2007) as well as to

systems which are composed of overlapping subsystems. The platoon of vehicles by Stanković, Stanojević, and Šiljak

Principle contributed by a clear conceptual insight and (2000) and Espinosa et al. (2007), and formation of aerial

effective solutions (İftar; 1993a,b; İftar & Özgüner, 1990; vehicles by Stipanović, Inalhan, Teo, and Tomlin (2004). An

Ikeda & Šiljak, 1986; Ikeda, Šiljak, & White, 1981, 1984; application to web handling systems was offered by

Šiljak, 1978, 1996). The structure of expansion–contraction Benlatreche, Knittel, and Ostertag (2005) and Sakamoto

relations including contractibility of controllers is analyzed and Kobayashi (2004), while Jung, Choi, and Seo (2000)

in Bakule, Rodellar, and Rossell (2000), Šiljak and have proposed an active suspension system for a car model.

Stipanović (2000), Stanković and Šiljak (2001), Stanković The superiority of decentralized control schemes over

et al. (2007), and Chu and Šiljak (2005) for LTI systems, centralized ones led to the concept of multiple control

while Bakule, Rodellar and Rossell (2002) and Stanković schemes when taking into account reliability issues of

controllers (Šiljak, 1991). The design of reliable controllers

is closely related to overlapping decompositions. New

applied results in reliable overlapping control are presented

in Bakule, Paulet-Crainiceanu, Rodellar, and Rossell (2005).

Delayed systems have been considered to solve the routing

problem in communication networks Ataslar andP Iftar (1999).

Bakule et al. (2005) developed a version of H 1 control for

continuous-time state delayed systems, while Bakule, Rodellar,

and Rossell (2006) present H 2 control for discrete-time state

delayed systems.

interconnected disjoint subsystems with a specific structure.

They are characterized by the identity of the subsystem

dynamics and the symmetry of the interconnections. The

symmetry gives the rise to essential simplifications of the

modeling, analysis, and synthesis. Particularly, the problem of

decentralized control design for the overall system can be

reformulated as a robust centralized control problem for a

reduced order design system. Thus, the complexity of the

design process is essentially reduced. H 2 control design of

delayed symmetric composite systems has been selected to

illustrate the main idea of this procedure.

Consider the symmetric composite system in the form:

þ ½B þ DBi ðtÞui ðtÞ þ si ðtÞ;

Fig. 4. Overlapping controller design: (a) overlapping subsystems; (b) yi ðtÞ ¼ ½C þ DCi xi ðtÞ; xi ðto Þ ¼ Fi ðto Þ;

expanded system; (c) decentralized controller design; (d) contracted closed-

loop system. d to 0; i ¼ 1; . . . ; N (24)

94 L. Bakule / Annual Reviews in Control 32 (2008) 87–98

Interconnections are considered as The systems (24) and (25) can be rewritten into the form:

X

N ẋðtÞ ¼ ½Ā þ DĀðtÞxðtÞ þ ½Ād þ DĀd ðtÞxðt dÞ

si ðtÞ ¼ ½Aq þ DAi j ðtÞx j ðtÞ þ ½Adq þ DAdi j ðtÞx j ðt dÞ

j¼1 þ ½B̄ þ DB̄ðtÞuðtÞ;

(25) yðtÞ ¼ ½C̄ þ DC̄ðtÞxðtÞ; xðto Þ ¼ Fo ðto Þ; d to 0

where the meaning of vectors and matrices in (24) and (25) (30)

remains the same as in (1) and (2) including the norm Now, it is sufficient to present the key relation only for the

bounded character of uncertainties. Notice that the nominal matrix Ā. We get

matrices A; B; C; Aq ; Adq are identical for all subsystems.

Constant matrices DðÞ ; EðÞ correspond with the uncertainty T T ĀT ¼ diagðAs ; :::;As ; Ac Þ (31)

matrix DðÞ analogously as in (3). They are identical for all

where

subsystems in (1) and (2). Uncertainties are lumped in

F ðÞ ðtÞ. As ¼ A A q ; Ac ¼ As þ NAq (32)

With the systems (24) and (25) we associate a quadratic cost:

when applying the transformation (29) on the system (30).

X

N N Z 1

X Analogous relations hold for the matrix Ād . It leads to two low

J¼ Ji ¼ ½xTi ðtÞQxi ðtÞ þ uTi ðtÞRui ðtÞ dt (26) order systems which describe the dynamic properties of the

i¼1 i¼1 0 original system. Thus, the original problem is reformulated as a

control problem for two simultaneous plants. The term NAq in

where Q and R are given positive definite matrices. (32) is decomposed into a nominal part Ao ¼ N=2Aq and an

The design objective is to find global decentralized resilient uncertain norm bounded part DAo ðtÞ ¼ Do F o ðtÞEo , where

dynamic controller quadratically stabilizing the system (24) Do Eo ¼ N=2Aq and F o ðtÞT F o ðtÞ 1. These manipulations

and (25) which guarantees the upper bound of the cost (26) for result in a control design model:

any admissible uncertainty. Suppose the decentralized dynamic

full order controller which is composed of N local feedback ẋr ðtÞ ¼ ½Ār þ DĀr ðtÞxr ðtÞ þ ½Ādr þ DĀdr ðtÞxr ðt dÞ

controllers:

þ ½B̄ þ DB̄ðtÞur ðtÞ; (33)

x̂˙ i ðtÞ ¼ ½Ac þ DAc ðtÞx̂i ðtÞ þ ½Bc þ DBc ðtÞyi ðtÞ; yr ðtÞ ¼ ½C̄ þ DC̄ðtÞxr ðtÞ

(27)

ui ðtÞ ¼ ½Cc þ DCc ðtÞx̂i ðtÞ; i ¼ 1; . . . ; N where

where x̂i ðtÞ is the controller state of the subsystem i. Ac ; Bc ; C c N

Ar ¼ A 1 Aq ; DAr ¼ DAðtÞ þ DAo ðtÞ (34)

are the controller matrices to be determined. These matrices are 2

identical for all subsystems.

To get a more insight into the structure of this class of with DAðtÞ ¼ DA F A ðtÞEA . Analogous relations hold for the

systems consider the transformation T of the state vector delayed terms in (33).

xðtÞ ¼ ½xT1 ðtÞ; . . . ; xTN ðtÞ

T A quadratic cost associated with the system (33) is

Z 1

x̃ðtÞ ¼ TxðtÞ (28) Jr ¼ ½xTr ðtÞQxr ðtÞ þ uTr ðtÞRur ðtÞ dt (35)

0

2 3

ðN 1ÞI I ... I I x̂˙ r ðtÞ ¼ ½Ac þ DAc ðtÞx̂r ðtÞ þ ½Bc þ DBc ðtÞyr ðtÞ;

6 I ðN 1ÞI ... I I 7 (36)

166 .. .. ..

7

.. 7; ur ðtÞ ¼ ½Cc þ DCc ðtÞx̂r ðtÞ

T¼ 6 . . } . . 7

N6 7

4 I I ... ðN 1ÞI I 5 which drives the system (33) optimally with a guaranteed cost

I I ... I I for any admissible uncertainty. This is a centralized control

2 3 design problem which can be solved by using the LMIs

I 0 ... 0 I (Mahmoud, 2004). A feasible solution of the problem (33)–

6 0 I ... 0 I7 (35) leads to the determination of the matrices Ac ; Bc ; C c and

6 . .. .. .. 7

1

T ¼6 . 6 . . } . .7 the upper bound on cost J o .

7

4 0 0 ... I I5 The implementation of the matrices Ac ; Bc ; Cc into the

I I . . . I I decentralized controller (27) leads to the basic result. The

(29) closed-loop system (24), (25), (27) is quadratically stable with a

L. Bakule / Annual Reviews in Control 32 (2008) 87–98 95

guaranteed upper bound cost J ¼ f ðNÞJ o , where the function 5. Decentralized networked control

f ðNÞ presents for instance (Yang & Zhang, 1995).

Delay free systems are considered in Hovd and Skogestad The current state of art as well as possible future trends in

(1994), Liu and Zhang (1999), Xiaoping (1992), Huang, Lam, technology advances in real world large scale and complex

Yang, and Zhang (1999), and Wang and Zhang (2000). systems, low cost processing, and communications lead to

Regional pole assignment by state feedback proposed Liu, Jing intensively increasing complexity of control systems. It

and Zheng (2006). Bakule and Rodellar (1996) and Yang and motivates the development of new theoretic approaches to

Zhang (1995) deal with the Ac ; Bc ; C c complexity reduced control systems. Two main changes in the control system

problem of decentralized stabilization. An extension to state- research directions are the explicit considerations of the inter-

dependent H 1 switched decentralized control design with connections and a renewed emphasis on distributed control

decentralized supervisor derived Bakule (2007). Decentralized systems which is closely related to decentralized control of

H 1 control and reliability issues developed Lam and Huang large-scale systems (Baillieul & Antsaklis, 2007).

(2007). Yan and Xie (2003) present a reduced-order control The notion of distributed control illustrates Fig. 5. Fig. 5(a)

design for a class of similar nonlinear systems. Real world shows decentralized controller in a BD form, while Fig. 5(b)

symmetric composite systems can be found in parallel systems presents a distributed controller where dashed lines correspond

such as flow splitting parallel reactors with combined with off-diagonal blocks given by communication links.

precooling Hovd and Skogestad (1994), electric power systems Recently, new methods and algorithms have been proposed to

operating in parallel Bakule and Lunze (1988) and Lunze include communication issues into the decentralized control

(1992), industrial manipulators with several degrees of freedom design framework. Such extensions concern the communication

Vukobratovic and Stokic (1982), flexible structures (Trächtler, among subsystems, local controllers, and communication in the

1991), space crystal furnace Ebert (1999), homogeneous feedback loop. Though a variety of structures and models in this

interconnected systems such as seismic cables El-Sayed and framework have been analyzed, there remains a gap between

Krishnaprasad (1981), or in various formation problems of decentralized control and control over networks. Current trends

vehicles in cyclic pursuit solved by using circulant matrices in in the use of networks for distributed control, diagnosis, and

Marshall, Broucke, and Francis (2004) and Roberts and Stilwell safety present Moyne and Tilbury (2007) with an emphasis on the

(2006). Steward, Gorinevsky, and Dumont (2003) present a network performance characteristics such as delay, delay

spatially distributed system applied to the paper machine variability, and determinism. Communication requirements for

problem. Other interesting results can be found in Hovd and decentralized control systems with noiseless digital channels and

Skogestad (1994) and Yang, Lam, and Zhang (1996). bounded system noise present Yüksel and Başar (2007) and

Delayed systems have been considered within the frame- Yüksel and Başar (2006). Two-agent optimal formation control

work of the guaranteed cost control design for state delayed problem with limited communication capacity consider Shi et al.

uncertain systems by Bakule (2005), while Bakule (2006) deals (2005). Decentralized synchronization of an uncertain time-

with the resilient stabilization by using the LMIs. varying network including several synchronization criteria are

given in Zhong, Dimirovski, and Zhao (2007). The dynamics of

local estimator-controller scheme with communication among

vehicles in cooperative formations are studied by Smith and

Hadaegh (2007). Hristu-Varsakelis (2005) surveys feedback-

based communication including a BD gain matrix for a

networked control system and the effects resulting from

unreliable communication links on the performance. Narendra,

Oleng, and Mukhopadhyay (2006) demonstrate through simula-

tion that the significant improvement in transient responses of the

disjoint subsystems can be achieved with communication at

relatively few instants of time. Roberts and Stilwell (2006) deal

with decentralized control and estimation with a circulant

communication network. Roberts and Stilwell (2005) developed

decentralized control over periodic fast switching network

applied to autonomous vehicle platoon with the network

characterized by using the graph theoretic issues. Dynamic

graphs present a new promising approach elevating the role of

interconnections on the same level as subsystems to shape the

performance of coupled systems (Šiljak, 2008). Rotkowitz and

Lall (2006) consider the problem of constructing decentralized

control minimizing a norm of the closed-loop subject to a

subspace constraint by using the concept of quadratic invariance.

Langbort, Gupta and Murray (2006) developed the LMI type

Fig. 5. Feedback structures: (a) decentralized control; (b) distributed control. condition for the existence of a decentralized controller for

96 L. Bakule / Annual Reviews in Control 32 (2008) 87–98

heterogenous interconnected subsystems with failing commu- Bakule, L., & Lunze, J. (1988). Decentralized design of feedback control for

large-scale systems. Kybernetika, 24(3–6), 1–100.

nication channels. Stubbs, Vladimerou, Fulford, Strick, and

Bakule, L., Paulet-Crainiceanu, F., Rodellar, J., & Rossell, J. M. (2005).

Dullerud (2006) describe the hovercraft flexible testbed for Overlapping reliable control for a cable-stayed bridge benchmark. IEEE

decentralized control of multiple autonomous vehicles that are Transactions on Control Systems Technology, 13(4), 663–669.

wirelessly networked and can be commanded from the Internet. Bakule, L., & Rodellar, J. (1995). Decentralized control and overlapping

Cantoni et al. (2007) present preliminary field trials of decomposition of mechanical systems. Part 1. System decomposition. Part

decentralized and distributed control design for the irrigation 2. Decentralized stabilization. International Journal of Control, 61(3),

559–587.

open-water channel. Bakule, L., & Rodellar, J. (1996). Decentralized control design of uncertain

Future research effort should consider the analysis and nominally linear symmetric composite systems. IEE Proceedings—Control

synthesis of complex strongly interconnected systems con- Theory and Applications, 143, 536–630.

trolled over realistic decentralized communication channels. Bakule, L., Rodellar, J., & Rossell, J. M. (2000a). Generalized selection of

complementary matrices in the inclusion principle. IEEE Transactions on

Performance of networked control systems and its limits are

Automatic Control, 45(6), 1237–1243.

key issues determined primarily by delays and dropped packets. Bakule, L., Rodellar, J., & Rossell, J. M. (2000b). Structure of expansion–

Reliability, sensor failures, actuator constraints as well as the contraction matrices in the inclusion principle for dynamic systems. SIAM

performance degradation under communication failures, recon- Journal on Matrix Analysis and Applications, 21(4), 1136–1155.

figurable control, and the emphasis on increased autonomy Bakule, L., Rodellar, J., & Rossell, J. M. (2002). Overlapping quadratic optimal

belong to other important current research challenges (Baillieul control of linear time-varying commutative systems. SIAM Journal on

Control and Optimization, 40(5), 1611–1627.

& Antsaklis, 2007). Bakule, L., Rodellar, J., & Rossell, J. M. (2005). Overlapping resilient H 1

control for uncertain time-delayed systems. In Proceedings of the 44th IEEE

6. Conclusion conference on decision and control and European control conference (CDC-

ECC’05) (pp. 2290–2295).

Bakule, L., Rodellar, J., & Rossell, J. M. (2006). Robust overlapping guaranteed

The paper presents past and present trends, and looks into cost control of uncertain state-delay discrete-time systems. IEEE Transac-

the future prospects in the area of decentralized control of tions on Automatic Control, 51(12), 1943–1950.

interconnected large scale and complex systems. The pre- Becerril, R., Aghdam, A. G., & Yurkevich, V. D. (2007). Decentralized two-

sentation has been focused on three important classes of time scale motions control based on generalized sampling. IET Control

decentralized control structures for strongly coupled complex Theory and Applications, 1(5), 1477–1486.

Befekadu, G. K., & Erlich, I. (2006). Robust decentralized controller

systems. An extension of decentralized control to networked design for power systems using matrix inequalities approach. In Proceed-

control systems is included. Due to our limitless desire to ings of the IEEE power engineering society general meeting. Paper

control larger and more complex systems, decentralized control 06GM0372.

remains a focal point of research in systems theory. Benlatreche, A., Knittel, D., & Ostertag, E. (2005). State feedback controllers

synthesis using BMI optimization for large scale web handling systems. In

Proceedings of the 16th IFAC world congress.

Acknowledgments Bernussou, J., & Sebe, N. (2002). About decentralized feedback: A survey. In

Proceedings of the 9th IFAC/IFORS/IMACS/IFIP symposium on large scale

This work was supported in part by the Academy of Sciences systems: Theory and applications (pp. 22–33).

Borrelli, F., Keviczky, T., Balas, G. J., Stewart, G., Fregene, K., & Godbole, D.

of the Czech Republic under Grant IAA200750802 and in part (2005). Hybrid decentralzied control of large scale systems. In Hybrid

by the Czech Ministry of Education, Youth, and Sports under systems: Computation and control (vol. 3414, pp. 168–183). Berlin/Heidel-

Grant LA 282. berg/New York: Springer.

Boyd, S., El Ghaoui, L., Feron, E., & Balakrishnan, V. (1994). Linear matrix

inequalities in system and control theory. Philadelphia: SIAM Studies in

References Applied Mathematics.

Cantoni, M., Weyer, E., Li, Y., Ooi, S. K., Mareels, I., & Ryan, M. (2007).

Aghdam, A. G., & Davison, E. J. (2007). Decentralized switching control for Control of large-scale irrigation networks. Proceedings of the IEEE, 95(1),

hierarchical systems. Automatica, 43, 1092–1100. 75–91.

Akar, M., & Özgüner, Ü. (2002). Decentralized sliding mode control design Chen, Y. H. (1989). Large-scale uncertain systems under insufficient decen-

using overlapping decompositions. Automatica, 38, 1713–1718. tralized controller. Transactins of the ASME, 111, 359–363.

Ataslar, B., & İftar, A. (1999). Decentralized routing controller design usign Chen, Y. H. (1992). Decentralized robust control design for large-scale uncer-

overlapping decompositions. International Journal of Control, 72(13), tain systems. The uncertainty is time-varying. Journal of the Franklin

1175–1192. Institute, 329, 25–36.

Aybar, A., & İ ftar, A. (2002). Overlapping decompositions and expansions of Chen, N., Gui, W., & Zhai, G. (in press). Design of robust decentralzied H 1

Petri nets. IEEE Transactions on Automatic Control, 47(3), 511–515. control for uncertain interconnected descriptor systems. Asian Journal of

Bakule, L. (2005). Complexity-reduced guaranteed cost control design for Control.

delayed uncertain symmetrically connected systems. In Proceedings of the Chen, Y. H., & Han, M. C. (1993). Decentralized control design for inter-

American control conference (pp. 2590–2595). connected uncertain systems. In C. T. Leondes (Ed.), Orlando, FL: Springer

Bakule, L. (2006). Resilient stabilization of uncertain state-delayed symmetric Verlag, Academic Press.

composite systems. In Proceedings of the 25th IASTED international Chen, N., Ikeda, M., & Gui, W. (2005). Design of robust H 1 control for

conference on modelling, identification, and control (pp. 149–154). interconnected systems. International Journal of Control, Automation, and

Baillieul, J., & Antsaklis, P. J. (2007). Control and communication challenges in Systems, 3, 143–151.

networked real-time systems. Proceedings of the IEEE, 95(1), 9–28. Chen, X. B., & Stanković, S. S. (2005). Decomposition and decentralized

Bakule, L. (2007). Stabilization of uncertain switched symmetric composite control of systems with multi-overlapping structure. Automatica, 41, 1765–

systems. Nonlinear Analysis: Hybrid Systems, 1(2), 188–197. 1772.

L. Bakule / Annual Reviews in Control 32 (2008) 87–98 97

Chen, X. B., & Stanković, S. S. (2007). Overlapping decentralized approach to Kown, O. M., & Park, J. H. (2006). Decentralized guaranteed cost control for

automation generation control of multi-area power systems. International uncertain large-scale systems using delayed feedback: LMI optimization

Journal of Control, 80(3), 386–402. approach. Journal of Optimization Theory and Applications, 129(3), 391–

Chu, D., & Šiljak, D. D. (2005). A canonical form for the inclusion principle 414.

of dynamic systems. SIAM Journal on Control and Optimization, 44(3), Kozáková, A., & Veselý, V. (2006). Robust decentralized controller design:

969–990. Independent design. In Proceedings of the 1st IFAC workshop on applica-

D’Andrea, R., & Dullerud, G. E. (2003). Distributed control design for spatially tions of large scale industrial systems.

interconnected systems. IEEE Transactions on Automatic Control, 48(9), Krishnamurthy, P., & Khorrami, F. (2003). Decentralized control and distur-

1478–1495. bance attenuation for large-scale nonlinear systems in generalized output–

Dullerud, G. E., & D’Andrea, R. (2004). Distributed control of heterogenous feedback canonical form. Automatica, 39, 1923–1933.

systems. IEEE Transactions on Automatic Control, 49, 2113–2128. Lam, J., & Huang, S. (2007). Decentralized H 1 control and reliability analysis

Ebert, W. (1999). Towards delta domain in predictive control—An application for symmetric composite systems with dynamic output feedback case.

to the space crystal furnace TITUS. In Proceedings of the 1999 interna- Dynamics of Continuous Discrete and Impulsive Systems, Series B: Appli-

tional conference on control applications (pp. 391–396). cations and Algorithms, 14(3), 445–462.

El-Sayed, M., & Krishnaprasad, P. S. (1981). Homogeneous interconnected Langbort, C. L., Chandra, R. S., & D’Andrea, R. (2004). Distributed control

systems: An example. IEEE Transactions on Automatic Control, 26, 894–901. design for systems interconnected over an arbitrary graph. IEEE Transac-

Espinosa, F., Awawdeh, A. M. H., Mazo, M., Rodriguez, J. M., Bocos, A., & tions on Automatic Control, 49(9), 1502–1519.

Manzano, M. (2007). Reduction of lateral and longitudial oscillations of Langbort, C., Gupta, V., & Murray, R. M. (2006). Distributed control over

vehicles platooning by means of decentralzied overlapping control. In Pro- failing channels. In P. J. Ansaklis, P. Tabuada (Eds.), Networked embedded

ceedings of the 46th IEEE conference on decision and control (pp. 690–695). sensing and control. Proceedings. LNCIS 331 (pp. 325–342).

Gajić, Z., & Ikeda, M. (2004). An overview of the collected works of Professor Lee, K. H. (2007). Robust decentralized stabilization of a class of linear

Dragoslav Šiljak. Dynamics of Continuous Discrete and Impulsive Systems, discrete-time systems with nonlinear interactions. International Journal

Series A: Mathematical Analysis, 11(2–3), 149–180. of Control, 80(10), 1544–1551.

Gajić, Z., & Shen, X. (1993). Parallel algorithms for optimal control of large Li, K., Kosmatopoulos, E. B., Ioannou, E. B., & Ryaciotaki-Boussalis, H.

scale systems. New York: Springer Verlag. (2000). Large segmented telescopes. Centralized, decentralized, and over-

Hovd, M., & Skogestad, S. (1994). Control of symmetrically interconnected lapping control designs. IEEE Control Systems Magazine, 20(5), 59–72.

plants. IEEE Transactions on Automatic Control, 30, 957–973. Liu, F., Jiang, P., Su, H., & Chu, J. (2002). Robust H 1 control for time-delay

Hristu-Varsakelis, D. (2005). Feedback control with communication con- systems with additive controller uncertainty. In Proceedings of the 4th world

straints. In D. Hristu-Varsakelis, W. S. Levine (Eds.), Handbook of net- congress on intelligent control and automation (pp. 1718–1722).

worked and embedded systems (pp. 575–599). Boston/Basel/Berlin: Liu, M., Jing, Y.-W., & Zheng, S.-Y. (2006). Pole assignment for uncertain

Birkhäser. symmetric circulant composite systems in a specific disk. IEE Proceed-

Huang, S., Lam, J., Yang, G. H., & Zhang, S. (1999). Fault tolerant decen- ings—Control Theory and Applications, 153(3), 357–363.

tralized H 1 control for symmetric composite systems. IEEE Transactions Liu, F., & Zhang, S. Y. (1999). Robust decentralized output feedback control of

on Automatic Control, 44, 2108–2114. similar composite systems with uncertainties unknown. In Proceedings of

İftar, A. (1993a). Decentralized estimation and control with overlapping input, the 1999 American control conference (pp. 3838–3842).

state, and output decomposition. Automatica, 29(2), 511–516. Liu, A.-J., Zhang, J.-F., & Jiang, Z.-P. (2007). Decentralized adaptive output–

İftar, A. (1933b). Overlapping decentralized dynamic optimal control. Inter- feedback stabilization for large-scale stochastic nonlinear systems. Auto-

national Journal of Control, 58(1), 187–209. matica, 43, 238–251.

İftar, A., & Davison, E. J. (2002). Decentralized control strategies for Lunze, J. (1992). Feedback control of large-scale systems. London: Prentice

dynamic routing. Optimal Control Applications and Methods, 23, 329–355. Hall.

İftar, A., & Özgüner, Ü. (1990). Contractible controller design and optimal Mahmoud, M. S. (2004). Resilient control of uncertain dynamical systems.

control with state and input inclusion. Automatica, 26(3), 593–597. Berlin: Springer Verlag.

İftar, A., & Özgüner, Ü. (1998). Overlapping decompositions, expansions, Marshall, J. A., Broucke, M. E., & Francis, B. A. (2004). Formations of vehicles

contractions, and stability of hybrid systems. IEEE Transactions on Auto- in cyclic pursuit. IEEE Transactions on Automatic Control, 49, 1963–

matic Control, 43(8), 1040–1055. 1974.

Ikeda, M., & Šiljak, D. D. (1986). Overlapping decentralized control with input, Moyne, J. R., & Tilbury, D. M. (2007). The emergence of industrial control

state, and output inclusion. Control-Theory and Advanced Technology, 2(2), networks for manufacturing control, diagnostics, and safety data. Proceed-

155–172. ings of the IEEE, 95(1), 29–47.

Ikeda, M., Šiljak, D. D., & White, D. E. (1981). Decentralized control with Mukaidani, H., Tanaka, Y., & Xu, H. (2003). An LMI approach to guaranteed

overlapping information sets. Journal of Optimization Theory and Applica- cost control of nonlinear large-scale uncertain delay systems under con-

tions, 34(2), 279–310. troller gain perturbations. Dynamics of Continuous Discrete and Impulsive

Ikeda, M., Šiljak, D. D., & White, D. E. (1984). An inclusion principle for dynamic Systems, Series B: Applications & Algorithms, S1, 40–45.

systems. IEEE Transactions on Automatic Control, 29(3), 244–249. Narendra, K. S., Oleng, N., & Mukhopadhyay, S. (2006). Decentralzied

Istepanian, R. S. H., & Whidborne, J. F. (2001). Digital controller implementa- adaptive control with partial communication. IEE Proceedings—Control

tion and fragility. New York: Springer Verlag. Theory and Applications, 153(5), 546–555.

Jamshidi, M. (1997). Large-scale systems: Modeling, control and fuzzy logic. Orqueda, O. A. A., Zhang, X. T., & Fierro, R. (2007). An output feedback

Upper Saddle River, New Jersey: Prentice Hall. nonlinear decentralzied controller for unmanned vehicle co-ordination.

Jiang, Z.-P. (2002). Decentralized disturbance attenuating output–feedback International Journal of Robust and Nonlinear Control, 17, 1106–

trackers for large-scale nonlinear systems. Automatica, 38, 1407–1415. 1128.

Jiang, Z.-P. (2003). Decentralized control for large-scale nonlinear systems: A Pagilla, P. R., & Zhou, Y. (2005). A decentralized output feedback controller for

review of recent results, Dynamics of Continuous Discrete and Impulsive a class of large-scale interconnected nonlinear systems. Journal of Dynamic

Systems. Series B S1. Applications and Algorithms, 38, 305–310. Systems Measurement and Control, Transactions of the ASME, 127,

Jung, Y. H., Choi, J. W., & Seo, Y. B. (2000). Overlapping decentralized EA 167–172.

control design for an active suspension system of a full car model. In Polendo, J., & Qian, Ch. (2007). Dcentralzied output feedback control of

Proceedings of the 39th SICE annual conference (pp. 85–90). interconnected systems with high-order nonlinearities. In Proceedings of

Kobayashi, Y., Ikeda, M., & Fujisaki, Y. (2007). Stability of large space the 2007 American control conference (pp. 1479–1484).

structures prserved under failures of local controllers. IEEE Transactions Richards, A., & Chow, J. P. (2007). Robust distributed model predictive control.

on Automatic Control, 52(2), 318–322. International Journal of Control, 80(9), 1517–1531.

98 L. Bakule / Annual Reviews in Control 32 (2008) 87–98

Roberts, D. G., & Stilwell, D. J. (2005). Control of an autonomous vehicle Šiljak, D. D., & Stipanović, D. M. (Stipanović, 2001). Organically-structured

platoon with a switched communication network. In Proceedings of the control. In Proceedings of the American control conference (pp. 2736–

American control conference (pp. 4333–4338). 2742).

Roberts, D. G., & Stilwell, D. J. (2006). Decentralized control and estimation for a Šiljak, D. D., & Zečević, A. I. (1999). In J. G. Webster (Ed.), Large scale and

platoon of autonomous vehicles with a circulant communication network. In decentralized systems (pp. 209–224). New York: John Wiley & Sons.

Proceedings of the American control conference (pp. 743–748). Šiljak, D. D., & Zečević, A. I. (2005). Control of large-scale systems: Beyond

Rosinová, D., & Veselý, V. (2006). Robust PID decentralized controller design decentralized feedback. Annual Reviews in Control, 29, 169–179.

using LMI. In Proceedings of the 1st IFAC workshop on applications of Vukobratovic, M., & Stokic, D. M. (1982). Control of manipulator robots:

large scale industrial systems (pp. 78–83). Theory and applications. Berlin: Springer Verlag.

Rotkowitz, M., & Lall, S. (2006). A characterization of convex problems in Wang, D., & Bao, P. (2000). Robust impluse control of uncertain singular

decentralized control. IEEE Transactions on Automatic Control, 51(2), systems by decentralized output feedback. IEEE Transactions on Automatic

274–286. Control, 45(1), 795–800.

Sakamoto, T., & Kobayashi, T. (2004). Decomposition and decentralized Wang, Y. H., & Zhang, S. Y. (2000). Robust control for nonlinear similar

controller design of web transfer system. Preprints of the 10th IFAC/- composite systems with uncertain parameter. IEE Proceedings—Control

IFORS/-IMACS/-IFIP symposium on large scale systems: Theory and Theory and Applications, 147, 80–84.

applications, vol. 1 (pp. 149–154). Wen, Ch. , & Zhou, J. (2007). Decentralized adaptive stabilization in the presence

Shi, L., Ko, Ch-K., Gayme, D., Gupta, V., Waydo, S., & Murray, R. M. (2005). of unknown backlash-like hysteresis. Automatica, 43, 426–440.

Decentralized control across bit-limited communication channels: Wu, H. (2003). Decentralzied adaptive robust control for a class of large scale

An example. In Proceedings of the American control conference (pp. systems with uncertainties in the interconnections. International Journal of

3348–3353). Control, 76(3), 253–265.

Shi, L., & Singh, S. K. (1992). Decentralized adaptive controller design for Xi, Z., & Ding, Z. (2007). Decentralized output regulation for large-scale

large-scale systems with higher order interconnections. IEEE Transactions output–feedback nonlinear systems with nonlinear exosystem. IET Control

on Automatic Control, 37, 1106–1118. Theory and Applications, 1(5), 1504–1511.

Smith, R. S., & Hadaegh, F. Y. (2007). Closed-loop dynamics of cooperative Xiaoping, L. (1992). Output regulation of strongly coupled symmetric compo-

vehicle formations with parallel estimators and communication. IEEE site systems. Automatica, 28, 1037–1041.

Transactions on Automatic Control, 52(8), 1404–1414. Yan, X.-G., & Xie, L. (2003). Reduced-order control for a class of nonlinear

Stanković, S. S., & Šiljak, D. D. (2001). Contractibility of overlapping similar interconnected systems with mismatched uncertainty. Automatica,

decentralized control. Systems & Control Letters, 44, 189–199. 39, 91–99.

Stanković, S. S., & Šiljak, D. D. (2003). Inclusion principle for linear Yang, G.-H., Lam, J., & Zhang, S. Y. (1996). Robust stability and stabilization

time-varying systems. SIAM Journal on Control and Optimization, of uncertain composite systems with circulant structures. In Proceedings of

42(1), 321–341. the 13th IFAC triennial world congress (pp. 67–72).

Stanković, S. S., Stanojević, M., & Šiljak, D. D. (2000). Decentralized over- Yang, G.-H., & Wang, J. L. (1999). Decentralized H 1 controller design for

lapping control of a platoon of vehicles. IEEE Transactions on Control composite systems: linear case. International Journal of Control, 72(9),

Systems Technology, 8(5), 816–832. 815–825.

Stanković, S. S., Stipanović, D. M., & Šiljak, D. D. (Šiljak, 2007). Decen- Yang, G.-H., & Zhang, S. Y. (1995). Stabilizing controllers for uncertain

tralized dynamic output feedback for robust stabilization of a class of symmetric composite systems. Automatica, 31, 337–340.

nonlinear intercionnected systems. Automatica, 43, 861–867. Yüksel, S., & Başar, T. (2006). On the absence of rate loss in decentralzied

Steward, G. E., Gorinevsky, D. M., & Dumont, G. A. (2003). Feedback controller sensor and controller structure for asymptotic stability. In Proceedings of the

design for a spatially-distributed systems: The paper machine problem. IEEE American control conference (pp. 5562–5567).

Transactions on Control Systems Technology, 11(5), 612–628. Yüksel, S., & Başar, T. (2007). Communication constraints for decentralzied

Stipanović, D. M., Inalhan, G., Teo, R., & Tomlin, C. J. (2004). Decentralized stabilizability with time-invariant policies. IEEE Transactions on Automatic

overlapping control of a formation of unmanned aerial vehicles. Automa- Control, 52(6), 1060–1066.

tica, 40, 1285–1296. Zečević, A. I., & Šiljak, D. D. (2004). Design of robust static output feedback

Stubbs, A., Vladimerou, V., Fulford, A., Strick, J., & Dullerud, G. E. (2006). A for large-scale systems. IEEE Transactions on Automatic Control, 49(11),

Hovercraft testbed for networked and decentralized control. IEEE Control 2040–2044.

Systems Magazine, 26(3), 56–69. Zečević, A. I., & Šiljak, D. D. (2005a). Global low-rank enhancement of

Swarnakar, A., Marquez, H. J., & Chen, T. (2007). A design framework for decentralized control for large-scale systems. IEEE Transactions on Auto-

overlapping controllers and its applications. In Proceedings of the 46th matic Control, 50(5), 740–744.

IEEE conference on decision and control (pp. 2809–2814). Zečević, A. I., & Šiljak, D. D. (2005b). A new approach to control design

Tamura, H., & Yoshikawa, T. (1990). Large-scale systems and decision theory. with overlapping information structure constraints. Automatica, 41, 265–

New York: Marcel Dekker. 272.

Tewarson, R. D. (1973). Sparse matrices. New York: Academic Press. Zhang, H., & Xie, L. (2007). Control, estimation of systems with input/output

Trächtler, A. (1991). Entwurf strukturbeschränkter Rückführungen an symme- delays. Berlin/Heidelberg: Springer Verlag.

trischen Systemen. Automatisierungstechnik, 39, 239–244. Zhong, W.-S., Dimirovski, G. M., & Zhao, J. (2007). Decentralzied synchro-

Vaz, A. F., & Davison, E. J. (1989). The structured robust decentralized nization of an uncertain complex dynamical network. In Proceedings of the

servomechanism problem for interconnected systems. Automatica, 25(2), 2007 American control conference (pp. 1437–1442).

267–272.

Šiljak, D. D. (1978). Large scale dynamic systems: Stability and structure. New

York: North Holland. Lubomı́r Bakule received the PhD degree in Control Engineering from the

Šiljak, D. D. (1991). Decentralized control of complex systems. New York: Czechoslovak Academy of Sciences, Prague, Czechoslovakia in 1974. Since

Academic Press. 1974 he has been with the Institute of Information Theory and Automation,

Šiljak, D. D. (1996). Decentralized control and computations: Status and Academy of Sciences of the Czech Republic, where he is a Director of

prospects. Annual Reviews in Control, 20, 131–141. Research. He has held numerous visiting appointments abroad. He has pub-

Šiljak, D. D. (2008). Dynamic graphs. Nonlinear analysis: Hybrid systems, 2, lished over 150 research papers and has led many research projects. His research

544–567. interests are in the theory and applications of robust decentralized control of

Šiljak, D. D., & Stipanović, D. M. (Stipanović, 2000). Robust stabilization of large-scale systems, networked control systems, delayed and switching control

nonlinear systems: The LMI approach. Mathematical Problems in Engi- systems, and structural control. He is Vice-Chair of the IFAC TC on Large Scale

neering, 6, 461–493. Complex Systems.

- EE4CL4_lecture30Transféré parMorteza Sepehran
- Control Final Exam Model. B..Transféré parEsmail M ALragami
- PID Tuning TutorTransféré parKiran Kulkarni
- Topic1 Control SystemTransféré parBrijbihari Singh
- State Feedback Control of Inverted PendulumTransféré parYANG PIN CHU
- Main Steam Turbine Controls Retrofit ISA FinalTransféré parvankayalasurya
- re30131_0203Transféré parWaheed Malik
- OGATA_docTransféré parYaaro
- Control and Simulation in LabVIEW 3-58 SMCIE TACN 2013Transféré parDaniela
- Dynamic Control of Electric Power SystemsTransféré parRahul Srivastava
- jgcdTransféré parjrbjrbFU
- Dynamic Control of Electric Power SystemsTransféré parMustafa Sarıkaya
- Solns Adaptive Control by AstromTransféré parAhmed H El Shaer
- Adaptive Control Design for High-Order MIMO Nonlinear Time-Delay Systems Based on Neural NetworkTransféré parivy_publisher
- 2013_2014_report1_copy (1)Transféré parChristian Gonzalez Aranda
- 0071462775_ar091.PDF Control Systems Based in Ogata BookTransféré parRaphael Almeida
- Introduction to ControlsTransféré parEng Mohammed
- Decentralized Design of Feedback Control of Large Scale SystemTransféré parBinhMinh Nguyen
- PidTransféré parLucas Matos Muniz
- Control System IntroductionTransféré parAnugrah Kesuma
- CHE 424 Process and Dynamics ControlTransféré parMaria Cecille Sarmiento Garcia
- 1-s2.0-S1474667016365612-main.pdfTransféré parsaeed
- QNET HVAC Workbook StudentTransféré parJoão Paulo
- Lecture_1 automaticsTransféré parali
- Tabuada Position Paper CPSTransféré pargiovanni
- Active ControlTransféré parJabez Richards
- 10.1.1.824.657Transféré parjae hak kim
- paperTransféré parbilal
- Application of fermentationTransféré parKaleem
- Experimental Procedure and ObservationsTransféré parwihanga

- Ljung - Asymptotic Behavior of EKF as a Parameter Estimator for Linear SystemsTransféré paraleong1
- CovertTransféré paraleong1
- Touchette - A basic introduction to large deviations.pdfTransféré paraleong1
- Candes, Li, Soltanolkotabi - Phase Retrieval via WIrtinger FlowTransféré paraleong1
- Spong, Block - The PendubotTransféré paraleong1
- Ljung, L. 2002. Prediction Error Estimation MethodsTransféré parFrancisco Calderón
- Bakule, Papik - Decentralized Control and CommunicationTransféré paraleong1
- 06824752.pdfTransféré parsherif_helmy
- OOP in Python-textbokTransféré paraleong1
- Python ScipyLectures-simple.pdfTransféré parArthur Vinícius
- Maddah-Ali, Niesen - Fundamental Limits of CachingTransféré paraleong1
- Lillicrap - Continuous Control With Deep RLTransféré paraleong1
- Reinforcement Learning in Robotics a SurveyTransféré paraleong1
- lecun2015Transféré parJohn Sun
- Rldm2015 Silver Reinforcement LearningTransféré parSharon Xu
- Arveson - ErrataTransféré paraleong1
- Atiyah, Macdonald SolutionsTransféré paraleong1
- moo_thesis.pdfTransféré paraleong1
- Opportunistc beamforming using dumb antennasTransféré paraleong1
- 620-332Transféré paraleong1
- Farber, Zeger - Quantization of Multiple Sources Using Integer BitTransféré paraleong1
- Zukerman - ClassnotesTransféré paraleong1

- Quadratic_Equations.pdfTransféré parlittlegus
- Helical GearTransféré parinduscad
- Expo PoisTransféré parDavey
- Elements of Mechanics of Elastic SolidsTransféré parNicolás Duque Gardeazábal
- Basic Analysis - K. KuttlerTransféré parrommel duran
- Plane Stress - Copy.pdfTransféré parJoão Santos
- ferTransféré parHicham Jad
- Tase Wire WarpingTransféré parAhmed Ramadan
- EE2202 Electromagnetic Theory Lecture NotesTransféré parkanjai
- 162234765 SPM Add Maths Formulae List Form4Transféré parAndrew Chew
- Tpde Imp Qustns Unit3Transféré parKrish Krishna
- Finite DiffTransféré parKumar Prashant
- NONINFORMATIVE BAYESIAN PRIORS. INTERPRETATION AND PROBLEMS WITH CONSTRUCTION AND APPLICATIONSTransféré parMahdi.Scribd.1
- Fine 1947 - Binomial coefficients modulo a primeTransféré parvojarufosi
- apostila-otimização-MITTransféré parMailson1
- Simple Harmonic Motion,linear motion,mechanics notes from A-level Maths TutorTransféré parA-level Maths Tutor
- KSR-Numerical MethodsTransféré parkpgs12
- practice_pre-calculus.docxTransféré parHector Dorian Alburquerque Morales
- Zill Calc CatalogTransféré parrrockel
- Root Locus ExamplesTransféré parahmed s. Nour
- 4. Extra Exercises[1]Transféré parmoiseicristian
- Sara Cuenda and Angel Sanchez- On the discrete Peyrard-Bishops model of DNA: Stationary solutions and stabilityTransféré parDopame
- GRE數學必備基本知識Transféré paran
- RusoTransféré parLeonardo Muñoz
- Higher Mathematics for Engineering and Technology Problems and SolutionsTransféré parAnonymous 5AamXv
- 6. NEXT CHAPTER MYP UP 9 Linear Relationships.docxTransféré parAmos D'Shalom Irush
- ApparentDip (1).xlsTransféré parSerge Cast
- RWGTransféré parPratik Chatterjee
- Quadratic Functions Maximum and MinimumTransféré parPerry Sin
- Aplications of Derivatives - ErrorsTransféré parSaahil Ledwani

## Bien plus que des documents.

Découvrez tout ce que Scribd a à offrir, dont les livres et les livres audio des principaux éditeurs.

Annulez à tout moment.