关键词: 复合故障, 因果链, 解耦, 故障诊断, 时间&mdash, 概率Petri网

Abstract: To analyze the composite fault in Discrete Event System(DES),a Time-Probability Petri Net(TPPN)and a decoupling algorithm were proposed.According to the observation event set and faults relationship set,an Extended Time Petri Net(ETPN)was constructed to describe the time sequence of cause-and-effect relationship,and ETPN was inversed and converted into TPPN,so that it could further represent the probability of event occurrence.An initial state was allocated to TPPN for its operation.The tokens were analyzed to find the fault origins when the TPPN was terminated,and the causal chains were extracted from ETPN.The simulation experiment showed that TPPN could accurately identify fault origins and decouple causal chains,even if the observation event set was incomplete with disturbance.Compared with other similar methods,TPPN had higher diagnosis precision.As a kind of DES model which could describe the timing sequence,probability and causal relationship effectively,TPPN could be further applied to behavior prediction and process analysis.

Key words: composite fault, causal chain, decoupling, fault diagnosis, time-probability Petri net