Զѧ
ACTA AUTOMATICA SINICA
2000Vol.26No.1P.74-78



ϵͳ̬Kalman˲㷨
÷
ժ Ҫ ӦִʱзܿصԻع黬ƽ(CARMA)Ϣģͣϵͳ̬Kalman˲㷨Riccati.Ϊ֤˲Ľȶԣѡ˲ֵʽ.˵㷨Ч.
ؼ ϵͳ̬Kalman˲㷨ȶԣִʱз.
NEW ALGORITHMS OF STEADYSTATE KALMAN FILTER
FOR STOCHASTIC CONTROL SYSTEMS
DENG Zili
(Institute of Applied Mathematics, Heilongjiang University, Harbin 150080)
LIU Yumei
(Department of Aviation, Civil Aviation Institute of China, Tianjin 300300)
Abstract Using the modern time series analysis method, based on the controlledautoregressive moving average(CARMA) innovation model, two new algorithms of steadystate Kalman filter gain for stochastic control systems are presented, where the solution of the Riccati equation is avoided. In order to ensure the as
ymptotic stability of the filter, two formulae of setting initial filtering estimate are given. A simulation example shows the effectiveness of the new algorithms.
Key words Stochastic control systems, algorithms of steady state Kalman filter gain, asymptotic stability, modern time series analysis.
1 
಻ͬſҪ̬Kalman˲1.̬Kalman˲Ҫ⸴ӵRiccati.Դˣִʱз2ݣCARMAϢģΪȡRiccatị˴̬Kalman˲㷨.ɢϵͳ
x(t+1)=x(t)+Bu(t)+w(t)(1)
y(t)=Hx(t)+v(t)(2)
״̬x(t)Rn۲y(t)Rmu(t)RP.
1. w(t)Rrv(t)Ǵֵİ
(3)
EΪѧΪתúţ
2. ϵͳȫɹ۵ģ
(4)
Цǿɹ۲ָ.(4)ʽȼڦȣ3ݣ
(5)
3 (u(t),u(t-1))֪ģu(t)Ϊƣu(t)(y(t),y(t-1))Ժ.
ǻ(y(t),y(t-1)u(t-1),u(t-2))x(t)̬Kalman˲
2CARMAϢģͺ
IiΪiiλʽ(1)ʽ(2)
y(t)=H(In-q-1)-1Bq-1,H(In-q-1)-1q-1ݣu(t),w(t)ݡ+v(t)(6)
q-1Ϊλͺ.طֽ
(7)
ʽĶʽXiΪϵnxΪףҽʽ(7)ʽ(6)CARMAϢģ
(8)
ȶģ4ݣD0=ImϢ(t)ֵΪQŵİ.D(q-1)QſGeversWouters2㷨.Ϊ֤D(q-1)ȶԣ׷ֽⶨ5ݣʽڵλԲϵʽ.
(t)(8)ʽȡֵ((t0-1),(t0-nd))ƼΪ
(9)
14. (10)
еʽ
(11)
.(12)
24. Nаֵ
(13)
36. ״̬x(t)зǵƱʽ
(14)
Цʽ(5)壬ҽα榸#=(䦸)-1ֿʾΪ
(15)
죬妷=-1T=--1B,L=--1
(16)
涨i=0(i<0)j0ͦiΪ
(17)
(18)
3 Ҫ
1. ϵͳ(1)(2)ڼ13£̬Kalman˲Ϊ
(19)
(20)
̬Kalman˲K㷨Ϊ
(21)
(22)
֤.ʽ(1)Ӱʽ2
(23)
(24)
(25)
ʽ(24)ʽ(23)ʽ(19).ʽ(13)ʽ(20).ʽ(14)(16)ʽ(25)1ʽ(21)(22).
ʽ(19)(9)ļ˲ֵϢֵ((t0-1)(t0-nd))й.֤ȶľֳֵǽȶģֳֵѡȡ޹أӶֳֵ.禵ǲȶʵѡ˲ֵϢֵǽȶ.
2. ڶ1£ӦK㷨ʽ(21)(22)ȡʽ(19)ĳֵ
Ϊ
(26)

(27)
ʽ(19)Ϣֵǽȶ.
֤.ף3ݵ֤.
עʽ(20)(23)(24)ͻɵõ̬KalmanԤ㷨Ƶߡ
4 
Ӧʽ(19)(20)(30)(31)ķͼ1ͼ2ʾ.

ͼ1 x1(t)̬Kalman˲

ͼ2 x2(t)̬Kalman˲
ϵͳ(1)(2)x(t)=x1(t),x2(t)ݡ䣬
(28)
v(t)=0.1w(t)+(t), u(t)=2sin(2t300).(29)
ʽw(t)ͦ(t)ֵΪĶ˹.֪S=0.1.CARMAϢģΪ
(30)
Ϣʽ(21)(26)
(31)
ȻѧĿ(69774019).
 19381962ҵںѧѧϵΪѧӦѧо.ҪоΪ״̬ƣ˲źŴ.150ƪ.
÷ 19711997ںѧԶۼӦרҵ˶ʿѧλΪйѧԺϵʦҪоΪ״̬ƣKalman˲.
(ѧӦѧо  150080)
÷(йúѧԺϵ  300300)
ο
1Vu K M. Discrete optimal stochastic controllers and its forms. Int.J. Systems Science,199021(3)567577
2һ.ִʱзӦ.֪ʶ磬19895470
3÷.һ̬Kalman˲佥ȶ.Ϣƣ199827(1)2631
4Deng Z L, Zhang H S et al.Optimal and selftuning white noise estimators with applications to deconvolution and filtering problems. Automatica, 199632(2)199216
5Newmann M M, Roberts A P. Polynomial optimization of stochastic discrete time
control for unstable plants.Int. J. Control,199651(6)13631386
6Ƽ.ǵ״̬Ƶļͳһ㷨.1998йѧļ.´ѧ.1998250254
ո 1997-09-29޸ 1998-07-25
