Զѧ
ACTA AUTOMATICA SINICA
1998 24 4 Vol.24 No.4 1998




ֱӱʶŶģ͵ģӦ
ƽ
ժҪֱڱʶйŶģͺ͹ģͲ㷨ŶģڼӦƣڲģԭδ֪ȷŶ.ʹ볣淽ͬĹ۲ʡȥ˷Ŷģ͵ļ.ַڶŶƵʵұʶĽʶ.˲ĿɱʶԺϵͳȶ.
ؼʡʶȷŶãӦ.
ADAPTIVE POLE PLACEMENT CONTROLLER WITH
DIRECTLY ESTIMATING DISTURBANCE MODEL

LIU HEPING
(Beijing University of Science & Technology, Beijing 100083)
AbstractA new algorithm which directly identifies parameters o f disturbance model and plant model in an estimator is given. The disturbance mo del is used for adaptive pole placement controller based on the internal model principle so that the disturbances can be eliminated. Because the identifing way is different from commonly used ones, it is not needed to separate the disturbance model . The method can be used in the situation where the disturbances are of many frequencies, and the degree of the estimator is not higher than that of the common one. The persisten t spanning of observing signal and system stability analysis are also given.
Key wordsIdentification,deterministic disturbances, pole placement , adaptive control.
1ǰ
ڼӦУδ֪ȷŶӰһֱǱĵĿ.رҲ ŶŶģͲͿƶڱʶϵĸԣһЩ㷨Ŷģ֪Ϊǰģ1.Щ꣬һЩ׶ģδ֪Ҳо˽ŶģзķڸĽȡŶģͷŬʹģͷ򻯣2ݣPalaniswami3÷ Ŷģ͹ֵֵɹ۲ֱӱʶ.ַ˱ʶ ĸ˷ģ͵ļ㣬ûиضڶסŶƵʶһ ʱĽ취.
ַȻ㷽죬ģͶǱ.ϵͳŶƵʵ  ӴļѶ.һֱܹӱʶŶģͼƶ ļӦ㷨ˣģз.Ӷ˼.ŶƵ ʿǶֵ.
2ܿϵͳ
躬ȷŶĿƶģͿвַ
A(z1)y(t)=zdB(z1)u(t)+d(t).(1)
ʽзֱʾtʱϵͳ롢ź.Z1Ϊһӳӣd(t)ΪֶнȷŶźţȡ
(2)
УdegD(z1)=2n.
ݼÿƵҪ(1)ʽʾϵͳB(z1)A(z1)D(z1)ʣӳʱd1Ϊ֪n,m,n֪.
3ŶģͼϵͳĹƷ
(1)ʽϵͳķǽдʽ
A(z1)y(t)=zdB(z1)u(t).(3)
ʽA(z1)=A(z1)D(z1),B(z1)=B(z1 )D(z1).ɴģ͹ƳA(t,z1)B(t,z1)Ϊ˹ŶļÿҪB(t,z1)ΪB(t,z1)D(t,z1).Ĳȡ´ʩʡȥ.ȣŶģдʽ
(4)
mn(4)ʽB(z1)D(z1)¹죬ʹ(1)ʽõµĲṹ
(5)

(6)
(5)(6)ʽõ

ʵıʶɵõģ͵ĹA(t,z1),B(t,z1),D(t,z1).
ڱʶЦ1(t-1)(3)ʽӦĦ(t-1)вͬʽͬźźͿź϶ɵģҪǷܹά릵(t-1)ͬıʶ.
1.PάźŦ (k)ɵ ϣ(k),N=(k),(k+1 )(k+N-1)нNеkųɵռάPΪP.S.(Persistently Spanning)ԣ4.P.S.Ա֤˿ɱʶ.
1. ʶ(7)ʽеĹ۲1(t-1)볣Ĺ۲(t-1)ͬP.S.ԵĳҪminm,n1.
֤..覵(t-1)P.S.ԣ
rank(t-1),N=n+m+1+4n=dim,
1(t-1),Nɱʾɣ(t-1)Nı任ʽ
1(t-1)N=(t-1)N.(8)



Rm+n+1+(3+m)nأݡ(n+m+1+4n)ϽΪ(n+2n)׵λ󣬼ΪI·Ϊ(m+1)(n+1)(2n+m+1)׾󣬼Ϊ1.m=1ʱ1(2+2n)(2+2n)ȷ.m=0ʱ1Ϊ(n+1)(2n+1)Ⱦۺ֪ʵȱ任
rank1(t-1),N=dim1m 1.(9)
Ҫ.m1ʱΪȾrank =n+m+1+4nأɵ
rank1(t-1),N=n+m+1+4nأdim1m1(10)
֤m1ıҪ.n1Ľͬ.
֤.
1˵minm,n1ʱ1(t-1)릵(t -1)ͬP.S.ԣdim1dimȣ˵ʶĲڦ(t-1).

4Ӧ㷨ջϵͳ
ƶŶģͲδ֪ʱñʶĲʽȷtʱ̵Ŀź.
D(t,z1)L(t,z1)u(t)=H(t,z1)yr(t)-R(t,z1)y(t),(11)
A(t,z1)L(t,z1)+zdB(t,z1)R(t,z1)=T(z1).(12)
T(z1)ȶָջϵͳ.yr(t)ΪοźţH(t,z1)ǰòΪ̬.ο뽥Ϊֵʱ趨Ϊ
H(t,1)=T(1)/B(t,1).(13)
2.Ӧ㷨ʹϵͳ㣺 
1) 1(t-1)ޣt;
2) (1Ϊ)
3) 
֤.ɶ1Ľۿ֪ɦ1(t-1) ɵıʶάP.S.Բ䣬ˣæ(t-1)㷨 ͬĲʶ.(7)ʽĵһʽд
1(t)=A(t,z1)D(t,z1)y(t)-zdB(t,z1)D(t,z1)u(t),(14)
ùֵɽ(1)ʽʾΪ
A(t,z1)y(t)-zdB(t,z1)u(t)=d(t)+š1(t).(15)
Ƚ(14)(15)ʽ֪d(t)൱d(t)Ĺֵš1(t)൱δģ̬
D(t,z1)d(t)=0,(16)
1(t)=D(t,z1)š1(t).(17)
(15)ʽӦ(11)ʽɵõջϵͳٽ(16)ʽ룬
(18)
ǵ(17)ʽĹϵɲʶнԼyr(t),d(t)н磬T(z1)ȶy(t),u(t)ڦ1(t)н磬ݱʶʣ5

ף5ݵĻ6.2.1ɵó1)2)Ľ.(18)ɵ
T(z1)y(t)-yr(t)=L(t,z1)1(t)+zdB(t,z1)H(t,z1)-T(z1)yr(t).
(13)ʽʽұߵĵڶt޵ļ޵.T(z1)ȶ1(t)0ʶн磬У˼23).
֤.
5ֵ
ñĿ㷨һȶķСλϵͳ
(1.0-2.1z1+1.1z2)y(t)=z1(1.0+1.2z1)u(t)+d(t)
ֵ棬ϵͳŶģͰδ֪d(t)ΪŶȡֵT=1ʾͼ1ͼ2.ͼ1ΪŶƵʵһd(t)=0.1sint.t=65ʱƵʴӦ=0.5ͻɦ=0.3.ͼ2Ŷк2Ƶʣd(t)=0.1sin0.5t+0.1sin0.3t.ο÷ֵΪ1ڷ.t25ĳڶԿźŽu(t)0.3޷.ķԿϵͳŶ˺ǿ.Ƕ̬ԻǾ̬ԶȡЧ.


(a) ϵͳӦ(b) źŹ
ͼ1n=1ʱķ


(a)ϵͳӦ(b)źŹ

ͼ2n=2ʱķ
6
ڼӦУֱӶδ֪Ŷģͼֱбʶķɼģͷļ.ṩ㷨ڶδ֪ƵȷŶʶĽ ڳͱʶ.Ϊ˱ֱʶP.S.ԶһĴۣӦΪminm,n1.
ߵλƼѧԶϵ 100083
ο
1Janecki D. Direct Adaptive Pole Placement for Plants Having Purely De terministic Disturbances. IEEE,Trans.Automatic Control, 1987,AC32(2):187-189
2ƽһ.δ֪ȷŶļӦƣԶѧ199622(4):401-409
3Palaniswami M. Adaptive Internal Model for Disturbance Rejection and Control. IEE PROCEEDINGSD,1993,140(1):51-59
4Shinaka S, Tanaka K, Suzuki T. A Simble Order Determination Method Based on a Generalized Adaptive Law.int, J. of control,198541(4)10 37-1050
5Goodwin G C, Sin K S. Adaptive Filtering Prediction and Control. NJ:Prentice hall,1984
ոڡ19960718
