Solving Method of H-Infinity Model Matching Based on the Theory of the Model Reduction

TELKOMNIKA Indonesian Journal of Electrical Engineering
Vol. 13, No. 2, February 2015, pp. 282 ~ 286
DOI: 10.11591/telkomnika.v13i2.6886  282
Received October 20, 2014; Revised December 18, 2014; Accepted January 3, 2015
Solving Method of H-Infinity Model Matching Based on
the Theory of the Model Reduction
Li Minzhi, Cao Xinjun
The School of Electronics and Information Engineering, Lanzhou Jiaotong University, Lanzhou, China
*Corresponding author, email: sqlmz@sina.com
Abstract
People used to solve high-order H model matching based on H control theory, it is too
difficult. In this paper, we use model reduction theory to solve high-order H model matching problem, A
new method to solve H model matching problem based on the theory of the model reduction is
proposed． The simulation results show that the method has better applicability and can get the expected
performance．
Keywords: high-order model, reduction theory, H model matching
Copyright © 2015 Institute of Advanced Engineering and Science. All rights reserved.
1. Introduction
H-infinity ( H ) optimal control theory of linear systems is a new kind of design method
developed in the end of 1980, and is the very active frontier subject in current control theory. In
many control systems, in order to improve the steady and dynamic performance of system, the
appropriate correction device needs to be added in the system, making the output
characteristics of the system meet all of the demand for performance specifics. This is the
model matching problem. In solving the model matching problem, it is mostly solved by
converting to H standard control problem [1-2]. Chen Yongjin proposed a kind of upper bound
method of searching for multi-blocks of model matching [3]. Zhuge Hai proposed an
approximate method of imprecise model matching [4]. These methods are easy to be achieved
for general systems, but these methods are more complicated for high order system model.
Moore proposed the balance order reduction problem of system in 1981 [5], then the method is
improved constantly [6], and some new reduction algorithms were put forward [7-9].
Due to the high order problem of system model in H model matching, combining with
the model order reduction theory, H model matching resolving method is proposed based on
model reduction theory. The analysis and simulation show that the method has good matching
characteristics.
2. H Model Matching Problem
 P s
 K s
z
y u
w
Figure 1. Principle figure of H standard problem
In control system, many H optimization problems of different requirements can be
converted into H standard problem. As shown in Figure 1, w is the external input, z is control

TELKOMNIKA ISSN: 2302-4046 
Solving Method of H 
Model Matching Based on the Theory of the Model… (LI Minzhi)
283
output, and u is the control input, y is the output of measurement.  P s is the generalized
controlled object,  K s is designed controller.
State equation of the generalized object  P s is described as:
1 2x Ax B w B u   (1)
1 11 12z C x D w D u   (2)
2 21 22y C x D w D u   (3)
Transfer function is:
 
1 2
11 12
1 11 12
21 22
2 21 22
A B B
P P
P s C D D
P P
C D D
 
       
    
(4)
Using the linear fractional transformation (LFT), transfer function from w to z can be
described as:
   
1
11 12 22 21,lG F P K P P K I P K P

    (5)
The H standard control problem is for a regular controller K , making the closed-loop
of system stable, and  ,lF P K 
less than a given , 0  .
w z

1T
G
2T K
Figure 2. Matching principle figure of H standard control model
H standard control model matching is shown as fig.2. Using three transfer function
matrix series 1T , K , 2T to approach transfer function G , the approximation degree will be
measured by 1 2G T KT 
 . The generalized controlled object:
  1
2 0
G T
P s
T
 
  
 
(6)
The controller is:
K K  (7)
A measure of model matching degree can be expressed as: 1 2G T KT 
 . When 1T and
2T are reversible, then the expression of model matching measurement is: 1 1
1 2T GT K 

 . So
1 1
1 2
ˆG T GT 
 , rG K , then, solving problem of H model matching can be transformed into
solving the model reduction problems, making ˆ
rG G

 within a required range.

 ISSN: 2302-4046
TELKOMNIKA Vol. 13, No. 2, February 2015 : 282 – 286
284
To make  ˆ
A B
G s
C D
 
  
  
a balance achievement.
Definition 1. Controllability and observability Gram matrix of system  A B C D，，， are
defined separately as follows:
0
At T A t
P e BB e dt
 
  (8)
0
A t T At
Q e C Ce dt
 
  (9)
A denotes the transpose of matrix A . It can be seen that the two matrices are
symmetric positive semi-definite matrixes, which satisfy the Lyapunov equation below:
0AP PA BB    (10)
0QA A Q C C    (11)
Diagonalization of the matrix ,P Q , then:
 1 1
1 2 1( , , , , )k k nTPT T QT diag      

       (12)
Where 1 2 1 0k k n            .
The system  A B C D，，， and  can be separated into blocks:
 11 12 1
1 2
21 22 2
, ,
A A B
A B C C C
A A B
   
     
   
(13)
 1 2    (14)
Where ( ) ( )
1 2,k k n k n k
R R   
    .
Theorem 1 [6]. Given asymptotically stable minimum system ˆG has Lyapunov
equilibrium form as follows:
 
11 12 1
21 22 2
1 2
ˆ
A A B
A B
G s A A B
C D
C C D
 
   
    
    
 
(15)
And there are:
1 2( )P Q diag   ， (16)
Where 1 1( , )kdiag     , 2 1( , )k ndiag     .
Reduced order model  
11 1
1
r
A B
G s
C D
 
  
  
which is truncated is asymptotically stable and
minimum system, and meet:
     1
ˆ 2r k nG s G s  

    (17)
The reduced order model  rG s is the K in the matching model we are asking for.

TELKOMNIKA ISSN: 2302-4046 
Solving Method of H 
Model Matching Based on the Theory of the Model… (LI Minzhi)
285
3. Simulation Examples
The mathematical expressions for state equation model of DC motor drive system is
[10]:
4
0 0 0 0 0 0 0 0 1.4
0 100 0 0 0 0 0 0 0
130 0 100 0 0 0 0 0 0
0 100 0.44 0 0 0 0 0 0
0 200 0.88 11.76 100 0 0 0 0
0 0 0 0 0 100 0 0 1.4
0 0 0 0 100 10 0 0 0
0 0 0 0 294.1 29.41 19.61 149.3 0
27.56 0 0 0 0 0 0 1.045 10 6.667
A
 
  
 
 
 
   
 
 
 
 
  
 
   
 0 1 0 0 0 0 0 0 0T
B 
 130 0 0 0 0 0 0 0 0C 
0D 
As 1T I and 2T I , output image for H model matching of system is shown as Figure 3(a),
the model matching solution is:
  
  
2
2 2
152.9247 4.96 255.7 28050
19.47 141.7 36.75 659.7
s s s
K
s s s s
  

   
(a) (b)
Figure 3. Output image of H model matching
As 1
1
100
T
s


and 2
1
5
T
s


, step response for H model matching of the system is
shown as fig.3 (b), the model matching solution is:
   
    
2
2
12611.7073 3369 2528 7.705 1639
158.1 41.78 7.206 27.46 334.3
s s s s
K
s s s s s
    

    
From the step response image of H model matching, it can be seen that the matching
model got by order reduction method and the step response of the original system are
completely consistent.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
50
100
150
200
250
300
350
Step Response
Time (sec)
Amplitude
原系统
模型匹配系统
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
50
100
150
200
250
300
350
Step Response
Time (sec)
Amplitude
原系统
模型匹配系统

 ISSN: 2302-4046
TELKOMNIKA Vol. 13, No. 2, February 2015 : 282 – 286
286
4. Conclusion
Using the principle of model order reduction to solve H model matching, from the
step response curve, it can be seen that the system has good tracking ability. The controller got
by this designing method has a certain practical application value, and model matching problem
of high order system will be solved well.
References
[1] Yuan SZ. Design of propulsion only emergency flight control system using H model matching.
Flight Dynamics. 2001; 19(1): 85-88.
[2] Shao KY, Jing YW, Li YS, Huang WD. Robust control system based on model matching. Journal of
Daqing Petroleum Institute. 1999; 23(3): 35-37.
[3] Chen YJ, Zuo ZQ, Wen SH, Ci CL. A solution of H control-model-matching problem. Journal of
Yanshan University. 2001; 25(z): 37-40.
[4] Zhuge H．An inexact model matching approach and its applications. The Journal of Systems and
Software (S0164-1212). 2003; 67(3): 201-212.
[5] Moore BC. Principal component analysis in linear systems: Controllability, observalility and model
reduction. IEEE Trans Automatic Control. 1981; ACO26(1): 17-31.
[6] K Zhou, JC Doyle, K Glover. Robust and optimal control. New Jersey: Prentice-Hall. 1996.
[7] Wang G, Sreeram V, Liu W Q. Balanced performance preserving controller reduction. System &
control letters. 2002; 46: 99-110．
[8] Serkan Gugercin，Athanasios C.Antoulas．A Survey of Model Reduction by Balanced Truncation
and Some New Results. International Journal of Control. 2004: 77(8): 748-766.
[9] Wang G, Sreeram V, Liu WQ. Performance Preserving Controller Reduction via Additive Perturbation
of the Closed-Loop Transfer Function. IEEE Transactions on Automatic Control. 2001; 46(5): 771-
775．
[10] Xue DY. Design and analysis for feedback control system. Beijing: Tsinghua University Press, 2000.

Solving Method of H-Infinity Model Matching Based on the Theory of the Model Reduction

More Related Content

What's hot (15)

Similar to Solving Method of H-Infinity Model Matching Based on the Theory of the Model Reduction (20)

More from Radita Apriana (20)

Recently uploaded (20)

Solving Method of H-Infinity Model Matching Based on the Theory of the Model Reduction