Զѧ
ACTA AUTOMATICA SINICA
1997 23 3 vol.23 No.3 1997



NARMAXģ͵
ARMAXģȫԻ
ء־
ժҪ˻NARMAXģ͵ķϵͳȫԻ.÷ʱARMAXģͽNARMAXģ.֤һԻнԣӦʵַ.˵˸÷Ч.
ؼʡϵͳNARMAXģͣARMAXģͣȫԻ.
THE GLOBAL LINEARIZATION OF NARMAX MODELS
VIA ARMAX MODELS
QIN BIN
(Dep of Automation, Shanghai Jiao Tong University, Shanghai 200030)
HAN ZHIGANG
(Institute of Applied Mathematics, Heilongjiang University, Harbin 150080)
AbstractIn this paper, a linearization method for complex and unknown nonlinear systems represented by NARMAX models is presented by using approximation ARMAX models. The limitation and construction method are discussed at the same time. The simulation result shows the efficiency of this method.
Key words Nonlinear systems, NARMAX models, ARMAX models, global linearization.
1
ģ֪δ֪ϵͳԻǽȻͬ.ģ֪ķϵͳԻģ12ݵ΢ּξȷԻӦԻģ3ݵĻ״̬۲Ի.ЩҪģͽṹ֪Ҫϸ.ڽ֪ģͽϽϵͳֵƽķ.ģ46ݲõǷֶԻķģ78ݲõķ.ģ9һִΧǶԻ˼룬֤˵뵥ϵͳǶģĴ.ģ10һԽǿľֲARMAXģԻ˸÷ĿԼн.ģ11ݵƽƽģ̬ARMAXģ͵ԻҲнǿ.һַϵͳȫԻ˼.һʱARMAXģͽNARMAXģͣ˹ѡȡ߾ȵķ.ڲʹ÷ھнǿ̬Եķϵͳ.
ϵͳNARMAXģLeontaritisBillings1985ģ12ݣָһķϵͳNARMAXģ
y(t)=f(y(t-1),y(t-p),u(t-1),,u(t-g),
e(t-1),,e(t-r))+e(t).(1)
e (t)ERmֵ y(t)YRm u(t)URnnm.
Ϊ۷㣬µm(p+r)+ngά
(t-1)=y(t-1)y(t-p),u(t-1),u(t-g),
e(t-1),e(t-r)ݦӣ
˦(t-1)ʦ=YpUgEr(1)ʽд
y(t)=f((t-1))+ e(t).(2)
Էģ(2)ɲúƽķõнģ.
2ϵͳȫԻ
Ϊ˵õ(2)ʽԶʽƽģͣȸһ
i:01ݣi=0
Цi()ȫΪ㣬աʦ.Եõĵʽ
(3)
һ뺯

Զ
(4)
f(i)Ϊf( )ڦiȡֵ.ȡ

(4)(2)ʽ֪()f()ÿiֵͬ.ˣĲֵ֪( )-f()ڦн.
1.13fڦn΢ģһСnĶʽP,ڦϵniֵf

С
2.ڷNARMAXģ(2)fһο΢ģΪʽ
(5)


(6)
֤.(4)ʽ1f()д
(7)
ȡf(i)Ϊf()ڦi̩չ10ݣ

ѡȡi=(t),֤.
3.ڷNARMAXģ(2)fһο΢ģԽƵ޸ARMAXģ
NARMAX(p,g,r)ARMAX(p,g,r,t).(8)
СաʾƵڣARMAX(p,g,r,t)ʾʱԻع黬ƽģ.
֤.Ϊ˱˵⣬Ե뵥ϵͳΪ.(5)(6)ʽֱӿɵ
(9)


 ֤.
3ȫԻнԼ
ɶ23֪(5)ʽԻ((t-1)).Ķ˵н.
4.(2)ʽķϵͳf((t))ÿ㴦p+1ο΢ģ
y (t)- (t)2M.
Ϊp̩չ.
֤.̩ն֪

NǦ((t-1))Ͻ.衬(t)-(t-1)p+12ϽΪ
֤.
ĶԵõ.
1.(2)ʽķϵͳf((t))(t)2ο΢ģ(6)ʽõ (t)ţ0KֻҪ(t)-(t-1)22K,Сy (t)- (t)2.
2.(2)ʽķϵͳf((t))(t)p+1ο΢ģСy (t)-p+1(t)2y (t)-p(t)2p(t)f((t))(t)p̩չ.
12ֱԻȵķС̩չ.Ȼ̩չ±ƽģ͵ĸӻͷԻ.ˣʵѡƵԵõԻģ;ָ壬ڹʵ.
4ϵͳȫԻʵ
(9)ʽ֪ϵͳ(1)Ϊ
 (t)=A(t) (t-1)+B(t)u (t-1)+C(t)e (t-1)+e (t).(10).
A(t)=A0(t)+A1(t)q-1+A2(t)q-2++Apq-pB(t)=B0(t)+B1(t)q-1+B2(t)q-2++Bgq-gC(t)=C0(t)+C1(t)q-1+C2(t)q-2++Crq-ry(t)YRmu(t)URne (t)ERmAi(t),Bi(t),Ci(t)ֱʱľ.
ǰ֪ӦԻARMAXģ͵Ĳδ֪.
(9)(10)ʽ֪
(11)
(t)=((t-1))+e(t).Խ(t)ɫ.(t)ƽȵARģͣ
(t)=P(t)(t).(12)
P(t)=1+1(t)q-12t)q-2+l(t)q-l(t).
(12)ʽ(11)ʽ
y (t)=A(t)y (t-1)+B(t)u (t-1)+C(t)e (t)+P(t)(t).
ʧһԣe (t)(t)أ
y (t)=A(t)y (t-1)+B(t)u (t-1)+C(t)e(t)+e(t).(13)
e(t)
C(t)=C0(t)+C1t)q-1Cs(t)q-s
(13)ʽеĲԲôӵĹС˷ʶ.ֵעǣȻ(13)ʽ(10)ʽʽͬģǴűʵ.ԭϵͳģʱԻģȻģ.
5
½Ϊ2ĵ뵥ϵͳ
y (t)=0.3y (t-1)u (t-1)+0.6y (t-2)
+2sin4(u (t-1))+cos(u (t-2))+e (t).
(13)ʽ֪ʽµARMAXģͱƽ
y (t)=a1(t)y (t-1)+a2(t)y (t-2)+b1(t)u (t-1)+b2(t)u (t-2)
+c1(t)e (t-1)+c2(t)e (t-2)+e (t).
ôӵĹС˷ɵýƵĹģ

ͼ1ͼ2ʾ.ɴ˿ɼиķǿеģн.

ͼ1y(t)(t)

ͼ2
6
ĸĻNARMAXģ͵Իнǿ.÷һ֡䡱ԻȽʺڸӵġģδ֪ķϵͳ绯̡̡ҩ̵.ڸ÷ֱӵõϵͳARMAXģͣڵõӦӦ.зͻŵ.ʱıʶһ׵£˸÷ľʵַдڽһо.
߼飺ء1966.1996񶫱ѧԿϵʿѧλ.ϺͨѧԶϵ²ʿо.ҪоΪϵͳıʶӦ³.
־ա1934.ѧڡѧְڣʿʦ.ҪоΪϵͳʶӦƣݽ׷ģͿ.
ߵλرϺͨѧԶϵϺ200030
־գѧӦѧо150080
ο
1Andrew T, Kadiyala R, Kokotovic P, Sastry S. Indirect techniques for adaptive input-output linearization of non-linear systems. Int. J. Control, 1991, 53(1):190-222.
2Sastry S , Isidori A. Adaptive control of nonlinear systems. IEEE Trans. Automatic Control, 1989, 34(11):1123-1131.
3Dawson D M, Qu Z, Carroll J C.On the state observer and output feedback problem for nonlinear uncertain dynamic system. Systems & Control Letters, 1992, 18(3):217-222.
4Skeppstedt A, Ljung L, Millnert M. Construction of composite models from observered data. Int. J.Control, 1992, 55(1):141-152.
5Billings S A , Voon W S F. Correlation based model validity test for nonlinear models. Int. J. Control, 1986, 44(2):235-244.
6Billings S A , Voon W S F.Piecewise linear identification of nonlinear systems. Int. J. Control, 1987, 46(2):215-235.
7Friedman J H. Multivariable adaptive regression splines. The Annal of Statistics, 1991, 19(1):1-141.
8Tolle H, Parks P C, Ersu E, Hormel M,Militzer J. Learning control with interpolating memories-general ideas, design lay-out, theoretical applicaches and pricatical applications. Int. J. Control, 1992, 56(2):291-317.
9ݺ־.ϵͳĽģԤ.Ӧã19863(2)64-71.
10Johanse T A , Bjarne A F. Contructing NARMAX models using ARMAX models. Int. J. Control, 1993, 58(7):1125-1153.
11ܳ.ϵͳĶģ̬ARMAXģ͡һֻڲֵ۵ģ.Զѧ199521(2)137144.
12Leontaritis I J, Billings S A. Input-output parametric models for nonlinear systems, Part deterministic nonlinear systems; Part: stachastic non-linear systems. Int. J. Control, 1985, 41(1):303-344.
13Cherey E W. Introduction to approximation theory, McGraw-Hill Book, Co., 1966.
ոڡ1994-12-07
