Զѧ
ACTA AUTOMATICA SINICA
1997 23 5 Vol.23 No.5 1997



³Ӧ㷨
Ͽ塡Ƽ
ժҪϵͳһ㲻ȷṹ£һֶӦ㷨˶Ӧ㷨⣬֤ȫԺȶ.㷨һƲԣ֤ģ͵һܿԣҵõܿضȵһ½.Ҫ֪ʶΪϵͳָָܹܿϽԼϽ.
ؼʡ³ӦãȷṹԣȫԺȶԣ
ROBUST MULTIVARIABLE ADAPTIVE POLE
PLACEMENT ALGORITHM
LI JUNMINGXING KEYIGAO SHUPING
(Department of Applied Mathematics,Xidian University, Xian710071)
AbstractAdaptive pole placement algorithm for multivariable system with a general uncertainty structure is presented.The long-standing adaptive pole placement problem for multivariable non minimum phase system with a general uncertainty structure is solved without a persistent excitation signal.A modification strategy for parameter estimation is proposed to guarantee the uniform controllability of the estimated model and to obtain a lower bound of the controllable measure.The global convergence and the stability of the algorithm are established.The scheme requires no a priori knowledge other than the system observability indices and upper bound of the controllability index.
Key wordsRobust adaptive pole placement,uncertainty structure,parameter estimate modification strategy,global convergence and stability.
1
Ӧ㷨³Դ19821ݱ󣬾һֱоſ֮һ޸ľӦ㷨(ģͶӰ)ʹĳ³Ҫо2.ԵȷϵͳR.Lozano3ȫӦ㷨½˾и³Ե㷨47ݣף7ݶԾвȷṹ׵ϵͳӦ.ڶȷϵͳף8оһӦ㷨ĿǾһ㲻ȷṹĶϵͳһֶӦ㷨һƲԣ֤ģ͵һܿԣõܿضȵһ½磬˶Ӧ㷨⣬֤ȫԺȶ.㷨ȲҪҲҪϵͳĳҪ֪ʶΪϵͳָָܹܿϽԼϽ.
2ģ㷨
Ķϵͳ
A(D)y(t)=B(D)u(t)+v(t),
(1)
СA(D)B(D)ֱqqqpάDĶʽ󣬲ʵ.A(D)мԼǦr[A(D)]ΪA(D)дϵ󣬲ʧһԣþΪԽԪΪ1Ǿ.y(t),u(t)ֱϵͳv(t)µĲȷ

(2)
(1)ʽɸдΪ·ʽ

(3)


ϵͳ.
¼裺
1)ϵͳ(1)ܿصģϵͳָܹvj(j=1,2,q)ָܿϽ֪.
2)Ͻiͦi,i=1,2,q֪.
弸ģ

(4)


ϵͳ(1)(3)Ĳδ֪ʱʱС㷨Ʋֵ

(6)

(5)
С>0ѡ񣬦i=1+tr(Fi(0)),tr(F(0))ЭļF(0)ѡ󣬦i(0)(i=1,q)ѡĲֵi(t)Ϊ*itʱֵ̹.
1.㷨(5)ʽԣ
1)Fi(t)ͦi(t)
2)
3)
(7)

֤.ף7ݵĶ1ֻ֤Ҫע±.
3Ƶ
(5)ʽƲδܱ֤ģ͵ĿɿԣģͲɿأ򼫵÷̳Ϊ췽̣¼㷨ʧ.(5)ʽһƵԣ.

(8)
Сi(t)(i=1,q)ѡȡںi(t)(i=1,q)Ҫʣ
P1)i(t)(i=1,q)
P2)i(t)(i=1,q)ʹSylvesterʽĳʽз½.
i(t)(i=1,q)ѡ񷽷¼Ǻţ

ɲĹSylvesterʽ£8ݣ

(9)
С
i(t),i=1,q,ѡΪ½ṹ

(10)
С(t)ڼDȡֵD=1,2,,lԪ

(11)
ǡ
ΪP2)(t)ֵDеһֵתΪһֵ8ݣΪP1)ڸıtʱһгͺȵĿغ.
1.=(t)ʱ

Ϊϵͳһֿɿض.
2.ͺ󿪹غ(t)£
(t)=(t-1)ЦjDzt(j)<(1+)zt((t-1)),
(t)=j,jzt(j)(1+)zt((t-1))zt(j)zt(i)iDС.(12)
2.ϵͳ(1)ʽ1)2)(5)ʽƣԾʣ
1)i(t),i=1,q
2)ϵͳĿɿض½

С0ǱϵͳĿɿضȵһ½磬

hi=max(1,hi0),i=1,2,,q.
֤.ȻЦi(t),i=1,qĶԪʽף8ݵ

(13)
Сni=ni-1+(n+pvi),n0=0,i=1,,q,g(i1,i2,,inq)ǦtFi(t)i=1,,qıgtǰпg(i1,i2,,inq)γɵv(t)ǶӦg(i1,i2,,inq)еɵ.
Ϊ˹(13)ʽv(t)°ţ
ȶЦi11,i1=0,1,,m1-1ɵȻйɵΪƵأйɵΪйɵΪˣйɵɶΪظейɵΪwnq彫(10)Цtõ

(14)
ڶԸȤԣǶÿDеԪγɵ
γһpt

(15)
NllάVandermonde.
ף7ݣõptһ½

(16)
֪

(17)
ɼ(14)(7)ʽ

(18)
hi=max(1,hi0),i=1,2,,q.(17)(18)ʽ

(19)
(19)ʽ(16)ʽ

(20)
ɶ1֪Fi(t)ͦi(t),i=1,2,qˣÿڼDȡֵiӦzt(i)Ҳ.zt()ͺ󿪹غ(12)еĵȱˣ(t)tҲ.
塡(12)(14)(15)ʽ

(21)
(20)(21)ʽ

(22)
ڿغͺˣ(t)δzl((t))=zmax(t)治ʽ

(23)
(22)(23)ʽõĽ.
4Ӧ㷨
ĹģΪ

(24)
ǵi.
1.(24)ʽеeai(t)治ʽ

(25)
С
֤.ף6ݵĽƹ㵽Σõ֤.
ڹģ(24)ʽĿ

(26)
Ц(t)Ƿ״̬Q*(D)ȶʽԽԪΪһ(-1)DĶʽ.y*(t)ǲοK(t,D),H(t,D)漫÷

(27)
СA*(D)ȶʽҦr(A*(D))=r[A*(D)],(24)(26)(27)ʽõջ̣89

(28)
3.ϵͳ(1)ʽδ֪(5)ʽƣñĹ޸ĹֵӦ㷨ʣһiķϽ磬ʹöһ(2)ʽиŵϵͳн.
֤.ף8ݣ9ݿ֪(28)ʽеĦǲt-t-1ĺˣtʱ.ԣt㹻ʱ(28)ʽӽ½ȶϵͳ

(29)
ڵн֪(t)ڡea(t).(26)ʽ֪y(t),u(t)Ҳڡea(t).ٶһl(t-1)Ϊ޽磬u(t),y(t)ea1(t)ͬ׷ɢˣһ½.l(t-1)Ϊ޽磬(4)ʽеĦl(t)ڦl(6)ʽеĽڦ1(21+Ŧl).ˣһ㹻СĽʹö½㷨(25)ʽìܣԣĽ۳.
5
Ծһ㲻ȷṹĶϵͳӦ⣬㷨ȫԡBIBOȶԺ³.һƵȷģ͵һܿԣһܿضȵһ½.
ߵλӿƼѧӦѧϵ710071
߼飺
1965.1989ҵӿƼѧ˶ʿѧλ.Ϊͨѧϵͳоְʿ.оӦƣԶ̬ϵͳŻƺܿ.
Ͽ塡19571994ҵͨѧϵͳоʿѧλ.ΪӿƼѧ.оɢ¼̬ϵͳۣPETRIۼӦ.
Ƽ1963.1995ҵӿƼѧ˶ʿѧλ.ʦоΪ³Ӧ.
ο
[1ݡRohrs C E et al.Robustness of Adaptive control algorithm in the presence of unmodeled dynamics.In:Proceedings of the 21st IEEE CDC,Orlando,Florida,1982:311.
[2]Ԭ.Ӧ³Եһ۲.Ϣƣ198917(4):3237.
[3]Lozano R,Goodwin G C.Globally convergence adaptive pole placement algorithm without a persistent excitement requirement.IEEE Trans on AC,1985,30(8):795798.
[4]Lozano R.Robustness adaptive regulation without persistency excitation.IEEE Trans on AC,1989,34(12):1260_1267.
[5]Lozano R,Dion J M,Dugard L.Singularity-free adaptive pole placement using periodic controllers.IEEE Trans on AC,1993,38(1):104108.
[6]Lozano R.Singularity-free adaptive pole placement without resorting to persistensy of excitation detailed analysis for first order systems.Automatica,1992,28(1):2733.
[7]Lozano R,Zhao X H.Adaptive pole placement without excitation probing signals.IEEE Trans,on AC,1994,39(1):112120.
[8].ȫĶӦ㷨.ߣ19916(4)277282.
[9]Goodwin G C,Sun K.Adaptive Filtering,Predition and Control,Englewood Cliffs,NJ:Prentice-Hall,1984.
ոڡ1995-08-25
