Զѧ
ACTA AUTOMATICA SINICA
1997ꡡ234ڡVol.23No.41997



ϸϢĸӶ̬
ϵͳιϼ﷽
ͮ˽
ժҪڶϵͳϵĹĿǺͶϵͳģ̬ľȷ壬һֹϵͳõĲģ͵ķɴһڴģвεƽ̵ķֶβde kleerGDEͨ. εϷʽϵЧʣ ϵͳģ̬ʵ屣֤˲μ޷㷨ڲͬμܵظ ˼Ĳμʴݹʽ. 
ؼʡϼϣ ܲģͣ μ﷽.
A HIERARCHICAL FAULT DETECTION AND DIAGNOSIS 
METHOD WITH PROBABILITY INFORMATION FOR 
COMPLEX DYNAMIC SYSTEMS
GE TONG
(Shanghai Jiao Tong University, Shanghai 200030)
DENG JIANHUA
(Northwestern Polytechnical University, Xian 710072)
AbstractFirst, a method to construct hierarchical models of a system for fault detection and diagnosis (FDD) is presented, based on consideration of the function feature of faults and accurate definition of the conception of mode. Second, a hierarchical FDD strategy is also presented, which uses the constructed models as fundamental descriptions of the system, and de kleer's general diagnosis engine (GDE) as the fundamental FDD tool. The hierarchical scheme can greatly improve diagnosis efficiency. Accurate definition of mode makes the system's hierarchies clear, prevents the algorithm from redundancy and produces compact probability transmission formulas.
Key wordsFault detection and diagnosis, function-oriented hierarchical model, hierarchical FDD method.
1
ǲϷ(J. de kleerGDEͨϵۣ12) һԪĸӶ̬ϵͳʱЧʺܵ. ڴϵͳ òϷЧʣιϵͳĲģԼôģƽ̵. ֻкٵ漰Щ34ݣ Ƕϵͳģ͵Ĺ춼ûȷ׼ģͲǺ. ⲻʹ﹤ڸϿܳظҲʹϢڲͬμļഫʵ. ΪˣĴӹϵĹϵͳģ̬ϸ壬 һϸϵͳģ͹췽ôģͲƽ̵ķ ڽЩ.
2ܲģ
ϵͳ(OBJʾ)СɵԪΪԪǿ滻СԪ. ɴ˿ɶϵͳĸ. 
1.(1)Ԫϵͳ(2) ϵͳɣһڲṹһܵOBJһϵͳ (3)ֻϵͳ. 
Ȼ ԪOBJϵͳ. 
ϵͳҪأն˱ڲṹ. ϵͳͨն˱OBJ໥ãն˱γϵͳı߽磬 ߽¼(ն˱ȡֵ)Ϊϵͳⲿ¼߽ԪļϳΪϵͳ. ϵͳɸͲεϵͳɣЩͲεϵͳΪ߲ϵͳ. COM(A)ʾAӦйϵCOM(A)иϵͳAһ. һʽ໥γɸ߲ϵͳ໥õķʽΪ߲ϵͳڲṹstru(.)ʾ. 
ϵͳĴΪĳЩܵģ ڲͬ¹ܲͬ. ɴ˿ϵͳģ̬ĸ. 
2.ϵͳAģ̬Mod(A)¼ϣ
(1) ԪA Mod(A)һ¼ռ䣬ÿһԪضӦAһֹ. 
(2) AԪҶAiCOM(A)Mod(Ai)Mod(A)={A(j)j=1,2,}ǵѿ˻Mod(Ai)(˼ҲΪһ¼ռ)һ֣ҶC1, CmA(j) (A(j)ΪMod(A)һԪ)Aⲿ¼
P(Cl)=P(Cm).
(1)
ʽP(.)Ϊʷ. 
Mod(A)ÿһԪسΪAһģ̬. COM1Ϊϵͳļϣôѿ˻ Mod(Ai)ΪCOM1ʵּΪINS (COM1) һԪسΪCOM1һʵ. ʵиԪصĴǲҪģ±ĽΪͨļ. 
϶彫OBJϵΪϵͳγӦģ̬γһֲνṹ ͲΪԪ߲㼴ΪOBJ. 
ϵͳܵģͳΪϵͳĹģͣ ǲͬģ̬ϵͳն˱ṩԼܺ. ͲϵͳĹģͰĳֽṹϳɸ߲ϵͳģͣģǺͽṹйصģΪװģ(device-oriented models) г γɸ߲ϵͳĹģ. mf(A)ʾAĹģͣ ms(A)ʾAװģͣͬһϵͳͲλֲͬжװģͣ ֱstru(ms(A))COM(ms(A))ʾAĶӦضװģms(A)ڲṹ ԡʾӳϵ ϹϵɱʾΪ

(2)
ms(A)mf(A).
(3)
 OBJһֶΡǶȵģ Ϊܲģ.
3ϼ
ϾʶϵͳǰԪʵּԪģ̬Ĺ. ǲεϷͬʱ޴ϵͳܵԪʵ֣ϷͨȷСΧֱԪķЧ. 
ùܲģͣʱiϵͳ̬άһOBJװģmi m0ӦOBJĹģ(ܲģ͵Ķ)ɿеװģ. mi+1ǽCOM(mi)ĳAjmf(Aj)ms(Aj)õģ AjΪmimi+1ķŴ mf(Aj)ms(Aj)ɹܲģṩ

(4)
Ȼ Ajѡ׼䴦ڹģ̬ĸ㹻󣬶COM(mi)Ԫؾʱmi+1Ϊmi. ϼԾͬһģmiеϻƺͲ{mii=0,1,}еģ͵ݽֹ. 
ϻƲͨGDEͨ2ݣԵǰģmiǰdi͵ǰѡCAΪ룬µĺѡCA. ѡCANiCAINS(COM(mi))ӼһԪسƺѡCANiеĺѡɽi-1ʱǰĲCAеĺѡɽiʱǰĲ. ʱiÿһѡӦһ. ںѡCi1CANiӦʼΪP(Ci1) ΪCi1. ںѡCi1CAӦʼΪP(Ci1) ΪCi1ĺ. ϻƸݵǰdiñҶ˹ʽP(Ci1). 
iʱ̣ģ͵ݽƳģmi(4)ʽmi+1⣬CACANi+1. Ci1CANiAjΪģmimi+1ķŴAjԪCi1аAjĳģ̬Aj(k), Aj(k)={Cp(Aj)p=1,2,,n},Cp(Aj)ΪCOM(Aj)nʵ. miչΪmi+1Ci1չΪCANi+1еnѡC(i+1)lp, p=1,2,,n,
C(i+1)1p=(Ci1-{Ai(k)})Cp(Aj).
(5)
ݶ2(1)ʽⲿ¼
P(Cp(Aj))=P(Cq(Aj))(Cp(Aj),Cq(Aj)Aj(k)),
(6)
ɴ˿ɵ
P(C(i+1)lpdi=vik)=P(Cildi=vik)P(Cp(Aj)Aj(k)), 
(7)
P(C(i+1)1pdi=vik)Ϊi+1ʱ̺ѡC(i+1)1pP(C(i+1)1p).
Ԫĳʼģ̬Ϊϵͳ֪ʶ. ļԪģ̬ת໥ Ϊͬһϵͳи򻥲ཻ ͬһиϵͳģ̬תҲ໥. ɹܲģ͵ĵײϣϵͳģ̬ʵֵĳʼ. 
4
ͼ1ϵͳOBJ ԪACTҺѹϵͳHYԴBAT1BAT2ԼSEN1SEN2SEN3(ֱҺѹp͵ѹv1v2) γϵͳADͼʾ. ϵͳģֿ̬½Уϵͳģ̬һӦܣ һӦϣ书δ֪ϵͳOBJAģ̬ӦDģ̬ӦBAT1BAT2. Ԫģ̬ʾΪ0.1 Ϲ1ʾѡģ̬Ӧϵͳ򵥱ʾÿһѡӦһʣֻʾǰѡ. Կ ϹûнϵͳDڲǲϷͨԪ. ʹڲη ûоȷʱеҲܲӦظ.


ͼ1ϵͳ
1Ϲ

iCANiCANiCOM(mi)
0(,0.65) (OBJ,0.34)(OBJ,1)(,0)OBJ
1(A,0.56) (D,0.27)(A,0.9)) (A,SEN2,0.05)A,D,SEN2
2(ACT,0.42) (HY,0.42)(HY,0.74) (ACT,HY,0.09)ACT,HY,SEN1,D,SEN2

ߵλ:ͮϺͨѧˮ¹оϺ200030
˽ҵѧ120䡡710072
ο
[1]de Kleer J, Williams B C. Diagnosing multiple faults. Artificial Intelligence, 1987, 32(1):97130.
[2]de Kleer J, Williams B C. Diagnosis with behavioral modes. In: Proc. IJCAI-89, 1989, 2:13241330.
[3]Ameen A H. Device understanding and modeling for diagnosis. IEEE Expert, 1991, 6(2):2632.
[4]Franz L, Wolfgang N. Integrating model-based monitoring and diagnosis of complex dynamic systems. In:Proc. IJCAI-91, 1991,2:10241030.
ոڡ1995-09-29
