Զѧ
ACTA AUTOMATICA SINICA
1997ꡡ234ڡVol.23No.41997



ɢϵͳ򼫵/Э
ȼʵһִ
Ӷ
ժҪɢϵͳģͼ·򼫵/Эȼʵַ.콵ģͣʹƥ򼫵̬Эֵ̬̬ܺҪڸģ. ֤ҪĽģ͵Ĵԣ ֱӸ˽ģ͵Ľʽṩ˵Եֵ. 
ؼʡɢʱϵͳģͼ򻯣ϵͳ . 
AN ALGEBRAIC APPROACH TO REGIONAL POLE AND COVARIANCE 
EQUIVALENT REALIZATIONS FOR DISCRETE-TIME SYSTEMS
WANG ZIDONG GUO ZHI
(Department of Automatic Control, Nanjing University of Science and Technology, Nanjing210094)
Abstract In this paper, a new approach for model reduction for discrete-time stochastic systems, which is called regional pole and covariance equivalent realization method, is presented. The major idea of the proposed approach is to construct the reduced-order models which match both the specified regional pole and steady-state covariance value. The resulting reduced-order models will approximate the full-order models in the two important aspects, i.e., transient and steady-state performances. The existence of desired reduced-order models is proved and the analytical expression of these models is also given. Finally, this paper provides an illustrative example.
Key words Discrete-time systems, model reduction, stochastic systems, approximation theory.
1Լ
ģͼһֱϵͳоΪԾ֮һַ. Ȼ๤ϵͳָ곣ֱӱΪϵͳ̬Эֵ ͳģͼ򻯷δҪϵͳ. ף12ͬʱƥq·Эģͼ򻯷q-Markov COVERȼʵַ÷Ҫԭϵͳ任Ϊɭ(Hessenberg)׼ΣӶʹ÷ѲӰ˽ģ͵Ĵ. һ棬ϵͳԼϵͳ̬ЭֵǷֱԶϵͳ̬̬ܵҪָ꣬ģͱ澡ܽӽԭģͣ͵򼫵/Эȼʵ(ΪRPCOVER's)оṩڹӦõģͼ򻯷. 
ȶԶɢϵͳ
x(k+1)=Ax(k)+Dv(k), y(k)=Cx(k).
(1)
xRnΪ״̬ yRm(mn)Ϊ. ADCΪάv(k)Ϊֵλǿȵĸ˹. ʼ״̬x(0)оֵ(0)ЭP(0) v(k). (A,D)ɿأ(AC)ɹ. 
֪ ϵͳ(1)̬ЭXE[[WTHX]x(k)xT(k)]ΪLyapunovX=AXAT+DDTΨһ⣬ЭΪ(Ai)T]CT. ٶϵͳ(1)λڵλԲڵԲڵq+joҰ뾶Ϊr0СԲڣСԲΪD(q,r). ĿǵRPCOVER'sΪm(mn)ɢϵͳ
xm(k+1)=Amxm(k)+Dmw(k).
(2)
(xnRm, w(k)Ϊλǿȵֵ˹)༴ȡAm,Dmʹ¼ָͬʱʵ֣1) ϵͳ(2)ļλD(q,r)ڣ 2) ϵͳ(2)̬ЭXm=E[[WTHX]xm(k)(k)]ǡΪY=CXCT.
2Ҫ֤
ڽ׼ģ͵Ĵ֤乹췽. ǳɢϵͳ(2)ЭXmһȼʵ֣ ҽAm,DmXm=AmXm+Dm.
13ݡ´󷽳
-qAmP-qP+AmP+(q2-r2)P=-Q.
(3)
Q0ΪAmļλڸԲD(q,r)ڣҽ(3)ʽP0. 
.ģ(1)D(q,r),ЭCXCT. ڵͽģ(2)伫λD(q,r)ڣ״̬ЭXm=CXCTҽP0, Q0
r2P-Q0,
(4)
Xm-(TVS-1+qI)Xm(TVS-1+qI)T0.
(5)
VΪ(VVT=I);S,TֱΪPr2P-Q0ƽ. һ(4)(5)㣬ͽģͲAm=TVS-1+qIDmΪ(5)ʽ˵ƽӣDm=[Xm-(TVS-1+qI)Xm(TVS-1+qI)T]1/2.
ҪԵ֤. ͽģ(2)1״̬ЭȼʵֵĶ֪شP0, Q0(3)ʽ
Xm-AmXm=Dm0.
(6)
ע⵽(3)ʽȼ(Am-qI)P(Am-qI)T=r2P-Q,ʽ˷ǸӶ(4)ʽȻһP=SST, r2P-Q=TTT,ʽΪ
[(Am-qI)S][(Am-qI)S]T=TTT.
(7)
ʽҽV4ݣʹ(Am-qI)S=TV Am=TVS-1+qI (6)ʽɵ(5)ʽ. 
Ե֤.ֻȡAm=TVS-1+qI,DmΪ(5)ʽ˵ƽӣS=P1/2, T=(r2P-Q)1/2, ׵. 
ע1. (4)(5)P0 Q0ش. ʵϣע⵽-1q1ʵѡȡPQʹTS-1Ԫ㹻СӶʹ(4)(5). ڼΣȡQ=r2P,ӶT=0,(5)ʽس. 
ע2.ע⵽򻯹д൱ɶ(ҪPQVѡȡ) µļָ꣬³ָ꣬H޷ָȣⷽодһ. 
3ֵ
Զɢϵͳ(1) 

ϵͳļΪ{0.578111,0.658489,0.575950.295573i,0.8074} ΪD(0.5,0.32)̬״̬ЭΪ

׼ģ(2)ʹAmļλD(0.5, 0.32)ڣ̬ЭXmΪģ͵ЭCXCT. Ͻṩķѡȡ

֤(4)(5)ʽ ɵļģͲֱΪ

ָ֤Լõ. ڸ߽׼ϵͳӦ߱Ƚͼƪȥ ˵зЧ. 
4
ɢϵͳģͼ򻯵򼫵/Эȼʵַּʹģ̬̬ҪƽģͣӶΪϵͳģͼṩһ;. һоڶָģͼϵĽ. 
1)ȻѧϾѧзչ. 
ߵλ:ϾѧԶϵϾ210094
ο
[1] Skelton R E, Collins E G Jr. Set of q-Markov covariance equivalent models of discrete systems. Int. J. Control, 1987, 46(1):112.
[2] Skelton R E, Anderson B D O. Weighted q-Markov covariance equivalent realizations. Int. J. Control, 1989, 49(5):17551771
[3] Furuta K, Kim S B. Pole assignment in a specified disk. IEEE Trans. Automat. Contr., 1987, 32(5):423427.
[4] Xu J-H, Skelton R E. An improved covariance assignment theory for discrete systems. IEEE Trans. Automat. Contr., 1992, 37(10):15881591
ոڡ1995-01-10
