Vous êtes sur la page 1sur 4

International Journal of Research in Management, Science & Technology (E-ISSN: 2321-3264)

Vol. 3, No. 2, April 2015


Available at www.ijrmst.org

A Review on Order Reduction of System using


Routh Approximation Method
Santosh Kumari#1, Ruchi Taneja#2
#

ECE Deptt. Prannath Parnami Institute of Management & Technology, Hisar, Haryana, India
1

santoshbhambu30@gmail.com
ruchitaneja.10@gmail.com

Abstract It is usually possible to describe the dynamics of


physical systems by a number of simultaneous linear differential
equations with constant coefficients. But for many processes (like
chemical plants and nuclear reactors etc) the order of the matrix
may be quite large. So it would be difficult to work with these
complex systems in their original form. The analysis and
synthesis of such higher order systems are difficult and generally
not desirable on economic and computational considerations. In
such cases, it is common to study the process by approximating it
to a simpler model. . For instance, the response of an airplane is
quite commonly approximated by a second order transfer
function. Through model order reduction, a complex higher
order system can be converted to a simple lower order system.
The reduced models are constructed such that all the parameters
can be preserved with acceptable accuracy.
Keywords accuracy, Integral Square Error, Model Order
Reduction, order of system, RAM, stability, transfer function

I. INTRODUCTION
In the design of control systems, it is often required to work
with mathematical models of high order. The analysis and
synthesis of higher order systems are difficult and generally
not desirable on economic and computational considerations.
Therefore, for computation, control, and other purposes, it is
often in demand adequately to comprise a high-order system
by a low-order model. The processes or devices can be
modelled by partial differential equations. To simulate such
models, spatial discretization via e.g. finite element
discretization is necessary, which results in a system of
ordinary differential equations, or differential algebraic
equations. After spatial discretization, the number of degrees
of freedom is usually very high. It is therefore very time
consuming to simulate such large-scale systems of ordinary
differential equations or differential algebraic equations.
A great number of problems are brought about by the
present day technology and societal and environmental
processes which are highly complex and 'large in dimension
and stochastic by nature'. The notion of 'large scale' is highly
subjective one in that one may ask: 'How large is large?'.
There has been no accepted definition for what constitutes a
large scale system. Many viewpoints have been presented on
this issue. One viewpoint has been that a system is considered
large scale if it can be decoupled or partitioned into a number
of interconnected subsystems or small scale systems for either

2321-3264/Copyright2015, IJRMST, April 2015

114

computational or practical reasons. Another view point is that


a system is large scale when its dimensions are so large that
conventional techniques of modeling, analysis, control, design
and computation fail to give reasonable solutions with
reasonable computational efforts. In other words a system is
large when it requires more than one controller. So, it is
highly desirable to obtain a reduced order model from the
complex higher order original system.
Through model order reduction, a small system with
reduced number of equations is derived. The reduced model is
simulated instead, and the solution of the original differential
equation can then be recovered from the solution of the
reduced model. As a result, the simulation time of the original
large-scale system can be shortened by several orders of
magnitude.
Model order reduction is a technique for reducing
the computational complexity of higher order mathematical
models in numerical simulations. Many modern mathematical
models of real-life processes pose challenges when used
in numerical simulations, due to complexity and large size.
Model Order Reduction aims to lower the computational
complexity of such problems, for example, in simulations of
large-scale dynamical systems and control systems. By a
reduction of the model's associated state space dimension
or degrees of freedom, an approximation to the original model
is computed. This reduced order model can then be evaluated
with lower accuracy but in significantly less time.
Goals of model order reduction are:

Simplifying the understanding of system for


design purpose.
To reduce the simulation time with the model of
system.
Reducing the controller size.
To economics in terms of hardware when
realizing the system.
Minimizing the computational complexity of the
model for system analysis.

Through model order reduction, a complex higher order


system is converted into a simplified lower order system. Let
the transfer function H(s) of higher order system is given by:

International Journal of Research in Management, Science & Technology (E-ISSN: 2321-3264)


Vol. 3, No. 2, April 2015
Available at www.ijrmst.org

b1Sn-1+......................+bn
H(s) =

bnSn-1+......................+b1

(1)

H(s) =

a0Sn+.......................+an
Let us assume that a reduced order model R(s) of order 'r'<n
which approximates the system H(s) is given by
b1Sr-1+.......................+br
R(s) =

(2)
a0Sr+.........................+ar

The coefficients of reduced order model are calculated such


that they good approximate the original system.
The rest of this paper is summarized as follows. Section II
describes background work related to various reduction
methods; an overview of Routh Approximation Method is
explained in section III, section IV describes performance
parameters for order reduction of the system, followed by
conclusion in Section V.
II. BACKGROUND WORK
Several model order reduction methods have been
anticipated by many authors for the time of last few decades.
A diversity of model order reduction techniques are used for
reduction purpose and they give diverse reduced model for
same system model but the value of a reduced model is finally
judged by the manner it is utilized. Some of the well known
methods are Pade Approximation Method, Routh
Approximation Method, Pole Placement Method, Particle
Swarm Optimization etc.
Routh approximation method is preferred over all the other
reduction methods because it is one of the most elegant
methods of model order reduction. It involves no Eigen value
evaluation and provides the capability to derive all stable
lower order models simultaneously via a single set of
computation. Due to the fact that it retains some of the initial
time moments only, the reduced model obtained by the RAM
tends to approximate the steady state response of the original
system in the time domain. Reduced system obtained by
Routh Approximation Method will always be stable provided
the original system is stable.

A linear time invariant system by m inputs and I outputs can


be symbolized by a matrix of transfer functions of the form (3)
through the numerator coefficients bi being I x m matrices. As
the denominator of a Routh approximant depends merely on
the denominator of H(s), a Routh approximant [1], [3] to an Ioutput, m-input system is computed by computing the Routh
approximant for every one term in the matrix of transfer
functions. The denominator of the Routh approximant merely
has to be computed once since it is similar for every term.
B. Routh Approximation Algorithm
Following steps are used for order reduction of system by
Routh Approximation Method:
Step 1: Determine the reciprocal of the transfer function H(s).
b1Sn-1+......................+bn
(s) =

(4)
a0Sn+.......................+an

Step 2: Construct the - Routh tables corresponding to the


reciprocal of the transfer function H(s).
TABLE I
ALPHA ROUTH TABLE
0

=a
0

=a
2

=a

a = a

a
0

1 =

a = a
0

a
1

a =a
2

a
1

a
0
1

III. ROUTH APPROXIMATION METHOD


A. Routh Approximation Method

a
0

2 =

It is a method of model order reduction that is used first to


develop the well known Routh Table for the original system
and then to construct the approximant in such a manner that
the coefficients of its Routh Table agree, up to a given order,
with those of the original system.
Let us consider a linear, time-invariant SISO system having
the transfer function

2321-3264/Copyright2015, IJRMST, April 2015

(3)
anSn+.......................+a0

a =a
0

a
2

...........

Step 3: For a kth order reduced model use recursive formulae


in Eqn. (5 and 6) to find
Rk(s) = Pk(s) Qk(s)

115

International Journal of Research in Management, Science & Technology (E-ISSN: 2321-3264)


Vol. 3, No. 2, April 2015
Available at www.ijrmst.org

Step 4: Reverse the coefficients of Pk(s) and Qk(s) again to


find reciprocal Rk(s). This reciprocal Rk(s) is the desired
reduced order transfer function.
In step 3, Pk(s) and Qk(s) can be found by using the following
formula
Pk(s)= k Pk-1(s) + Pk-2(s) + k
Qk(s)= k Qk-1(s) + Qk-2(s)

(5)
(6)

=b
1

=b
2

=b

b = b

b
0
1

b = b
0

a
1

b =b
2

a
1

[1]

b
0

2 =

4
0

b =b
2

(8)

V.
CONCLUSION
A study of model order reduction to reduce the order of a
system using suitable reduction method, preserving stability
with acceptable accuracy of the reduced system is done. A
survey of some pre-existing model order reduction methods
concludes that any Routh approximant of a stable system is
stable. It not only preserves stability but also has other
interesting and useful properties. Since Routh approximation
method retains some of the initial time moments only so the
reduced model obtained by Routh Approximation Method
tends to approximate the steady state response of the original
system. This work can be extended to find some highly useful
algorithms for model order reduction which can be used to
improve the existing methods of model order reduction.

TABLE II
BETA ROUTH TABLE

(t) dt

It is assumed that H(s) is the transfer function of a stable


system.

With, Q-1(s)= 0 and P-1(s)= 0


Q0(s)= 1 and P0(s)= 0

(7)

B. Impulse Response Energy


Impulse response describes the reaction of the system as
a function of time. If h(t) is the impulse response of the
transfer function H(s) then the impulse response energy be
defined by the integral of impulse response and is given by

h2 =

k = 1, 2, 3, 4...

1 =

ISE =

a
2

...........

[2]

[3]

Advantages of Routh approximation method:


Stability of system is preserved, when the original
system is stable.
Poles and zeros of the reduced system approaches
poles and zeros of the original system as the order of
the approximation is improved.
Steady state of system is preserved.
Disadvantages of Routh approximation method:
Gives no guarantee that transient state of system is
preserved.
IV. PERFORMANCE MEASURES
A. Integral Square Error
Integral square error is formed by integrating the square of
the system error over a fixed interval of time; it is frequently
used in linear optimal control and estimation theory.

2321-3264/Copyright2015, IJRMST, April 2015

116

[4]

[5]

[6]

[7]
[8]

[9]

REFERENCES
M. F. Hutton and B. Friedland, Routh approximants for reducing
order of linear time-invariant systems, IEEE Trans, vol AC-20,
June 1975, p329-337.
Y. Shamash, Model reduction using the Routh stability criterion
and the Pade approximation technique", Int. J. Control, Vol. 21,
No. 3, pp. 475-484, 1975.
Singh v., "Improved stable approximants using the Routh array",
IEEE Trans., AC-26, pp. 581-583,1981.
Wan Bai-Wu., Linear Model reduction is Using Mihailov
Criterion and Pade Approximation Technique, International
Journal of Control, 33(6), 1073, 1981
Y. Smamash Truncation method of reduction a viable
alternative electronic letters 22nd January, 1981,Vol 17,No.2 pp
97-98.
H. Inooka, G. Obinata and M. Takeshima, Design of a digital
controller based on series expansions of pulse transfer functions,
Journal of Dynamic systems, Measurement and Control, Vol. 105,
No. 3, pp. 204- 207, 1983.
Aoki, M., Control of Large-Scale Dynamic Systems by
Aggregation, IEEE Trans. Autom. Control, 13, 246-253, 1986.
Bandyopadhyay, B., Ismail, O., and Gorez,R., Routh Pade
Approximation for Interval Systems, IEEE Trans. Autom. Control,
39, 24542456, 1994.
Hwang, C., and Yang, S.F., Comments on the Computation of
Interval Routh Approximants, IEEE Trans. Autom. Control, 44
(9), 17821787, 1999.

International Journal of Research in Management, Science & Technology (E-ISSN: 2321-3264)


Vol. 3, No. 2, April 2015
Available at www.ijrmst.org
[10] T.H.S. Abdelaziz and M. Vala sek,"Pole-placement for SISO
linear systems by state-derivative feedback", IEE Proc.-Control
Theory Appl., Vol. 151,No.4,July,2004.
[11] Parmar, G.A Mixed Method for Large-Scale Systems Modeling
Using Eigen Spectrum Analysis and Cauer Second Form,IETE
Journal of Research,53,2, 89-93,2007.
[12] Dingy Xue, YangQuan Chen, Derek P. Atherton, Linear
Feedback Control Analysis and design with MATLAB, 2007.

2321-3264/Copyright2015, IJRMST, April 2015

117

[13] Younseok Choo., A Note on Discrete Interval System Reduction


via Retention of Dominant Poles, International Journal of
Control, Automation, and System,5(2),208-211,2007.
[14] M Gopal Digital Control and State Variable Methods Third
Edition, New Delhi, 2009.
[15] Singh, V. P., and Chandra, D. Routh Approximation Based Model
Reduction Using Series Expansion of Interval Systems, IEEE
International conference on power, control & embedded systems
(ICPCES),1,1-4,
2010.

Vous aimerez peut-être aussi