Զѧ
ACTA AUTOMATICA SINICA
1997ꡡ234ڡVol.23No.41997



Ŷϵͳʱ
ELS ʶֵʵ1)

ժҪعŶϵͳʱ ELS(С)ʶֵֺͳ΢ֵַŷ(Euler)- (Runge-Kutta)ʱ ELS ֵʵַͷ.
ؼʡϵͳƣС˷΢ֵַ.
NUMERICAL REALIZATION OF THE CONTINUOUS-TIME
ELS IDENTIFICATION FOR CONTINUOUS SYSTEMS 
WITH CORRELATIVE DISTURBANCES
ZHAO MINGWANG
(Dept.of Automation,Wuhan Yejin University of Science & Technology,Wuhan 430081)
AbstractFirstly,the continuous-time ELS identificaiton method for stochastic continuous systems with correlative disturbances are reviewed.Then two numerical realization methods for the ELS method and their simulation results are given based on the numerical integral,and Euler method and Runge-Kutta method for ordinary differential equations.
Key wordsStochastic continuous systems,parameter estimation,least squares method,numerical solution of ordinary differential equation.
1ʱ ELS ʶ
 30 ɢϵͳʶۺӦ϶ȡ˷˶ɹϵͳʶѷɫö.ĿǰӦʱһ㽫ϵͳΪɢϵͳɢϵͳʶ.ִӲȶ㣬ʶЧ.ϵͳΪвɲŶͲоϵͳʶҪ. 10 ꣬ϵͳʶоȡһЩĽչ.磬º˱ʶϵͳģͺŶģ͵ʱ ELS 1.Gevers ˶ϵͳǶȶ˲һнƽʶ2.Sagara ˻ֵֽϵͳΪɢģͣغֵɢƽģ̻ɢϵͳʶʶϵͳ3.÷ȱǲܵõŶģ.45ݻбƽС˷ʶϵͳڱƽֵԷָС˱ʶƫһģƫһĸ Markov .
ģ1ݵ ELS ʶĿǰΪ̵ġǿһϵͳʶ.۵΢ַŶ SISO ϵͳıʶ
A(S)dy1=B(S)dut+C(S)dt, D(S)t=t.
(1)
 yt  ut Ϊɲ룻t  t ֱΪɲŶ WieneṛSΪ Ito ӣ A(S)=1+a1S++anaSma; B(S)=b1S++bnbSmb; C(S)=1+c1S++cncSmc.
ģ1ݼ D(S)=1+d1S++dndSnd֪ȶ˲΢ַ(1)Ľ.ʧһԣٶ ncna.ˣϵͳģ(1)ɱʾΪعģ

(2)

=c1-a1cna-anab1bnb-c1-cncݦӡci=0,inc.
ʽ(2)ʾعģͣģ1 ELS ʶ㷨

(3)
 
ELS ʶϵͳģ A(S)  B(S)ҲʶŶģ C(S).ģ1ݻһ.񣬶Ը ELS ʶоԷ乤ʵһֱȱо.ֵʵϵͳ ELS ʶ.
2ŷʱ ELS 
ϵͳĲΪ hֵΪ yk  ukǻ Snyt,Snut  Sn() kh ʱ̵ֱֵֵΪ,.ڿźŵĹ⻬Բѣɲֵֵʱѡü㡢ڵʵֵĸη.ɸηֵĵʽΪ

(4)
ƵؿɵõĵƼʽ.ֵֵµ㷨ĵֵ

(5)

ŷǼЧĳ΢ֵַⷨ6ݣĳ ELS.ŷ˼룬ɵϵͳ(1) ELS ʶŷʵ㷨

(6)
С
(7)
ʵ㷨ϵͳʶĵ㷨ܽ£
1.ȷϵͳģͽ״ na,nb  nc hԼʶֵ 0  P0
2.Բ h òֵ uk+1 yk+1;
3.Ƽͣʽ(5)㣬ʽ(7);
4.ʽ(6)Ƽ k+1 Pk+1.ز2һεֵ.
3-ʱ ELS 
-Ǹ߾ȵĳ΢ֵַⷨ6ݣ侫뷽к(ź)Ĺ⻬й.ǵϵͳźŹ⻬Եԭö- ELS ˵.ˣϵͳ(1) ELS ʶ-ֵʵ㷨

(8)

(9)

(10)

(11)
С
(12)
ƼУΪµ㷨е(n)ֵ

(13)
С
ʵ㷨ϵͳʶĵ㷨ܽ£
1.ȷϵͳģͽ״na,nb nc  hԼʶ㷨ֵ 0 P0
2.Բh òֵ uk+1 yk+1;
3.Ƽ,ʽ(5),Ȼʽ(7);
4.ʽ(9)Ƽ K,1 Kp,1,Ȼʽ(13) ,ݴʽ(12),ʽ(10)(11) K,2 Kp,2.
5.ʽ(8)ֵ k+1 Pk+1.ز2һεֵ.
4
1.˲ D(S)=1+0.5S Ϊ֪2ϵͳ
(1+3S+2S2)dyt=1.5 utdt+(1+0.3S)dt,D(S)dt=dt.
(14)
£1)ԲΪ 0.005 ֵģϵͳ(14)У ut Ϊsin0.7t+0.5 sin0.2t,Ŷ dt Ϊ-dt,dtĸ˹.2)Բ 0.08 ȡǰһģе ut  yt Ĳֵ.ڴʱΪģв 1.6 ʴ¿Ϊȡϵͳ.3)ûŷ- ELS ֵʵַʶϵͳ(14).
1ʾзʱΪ 80 s(ƴΪ1000).ɱ1֪Ļڳ΢ֵַ ELS ʶʮЧģ-ķ.ϵͳ ELS ʶһѾõ֤ʵַǾҪ.ĵĺ˲ D(s) δ֪ʱıʶ⣬ԼϵͳӦƷƹ.
1

ŷ-
a1a2b1c1a1a2b1c1
=0.13.01202.00741.49610.26843.00962.00691.49740.2691
=0.53.03141.97611.49170.22123.02772.01641.49210.2339
=1.03.04881.96371.48130.18543.03722.03181.48620.2075
桡ֵ3.02.01.50.33.02.01.50.3

1)ұҵо.
ߵλ人νƼѧԶϵ人430081
ο
1ݡChen H F,Guo L.Continuous-time stochastic adaptive tracking:robustness and asymptotic properties.SIAM J.control and Optimization,1990,28(3):513527.
2ݡGevers M,Goodwin G C,Wertz V.Continous-time stochastic adaptive tracking.SIAM J.Control and Optimization,1991,29(2):264282.
3ݡSagara S,Yang Z J,Wada K.Recursive identificaiton algorithms for continuous systems using an adaptive procedure.Int.J.Control,1991,53(2):391409.
4ݡZhao M W.Markov Parameter estimation for stochastic continuous systems via Chebyshev polynomials.Preprints of The 10th IFAC/IFORS Symp.on Syst.Iden.,Finland,1994.
5ݡ.عģͻ Laguerre ʽƽ Markov .ֵӦã199516(2)127133.
6ݡ.ֵ.人йѧԺ磬1983.
ոڡ1995-08-31
