Research of probability Petri nets model for fault diagnosis based on Bayesian theorem

In order to eliminate the influence of incomplete sample data in fault diagnosis, Bayesian theorem is used for the design of Petri nets diagnostic model. Firstly, the concept about probability Petri net is defined. It can calculate the occurred probability of the absent datum based on Bayesian theorem from the detected attributes. Secondly, for the complicated Petri nets will induce the combinatorial states expand quickly; the design of probability Petri nets is based on reduced sample set to abate the complexity of the model. In order to create the logical relation between condition attributes and the faults, Skowron default rule generation method is used to obtain diagnostic rules. And for the programmable design of the model, all diagnostic rules are transformed to a series of basic rules that is defined. Finally, the diagnostic process is discussed. The probability Petri nets model can reduce the general probability of faults based on the complemental information. And the logical consequence can be calculated by incidence matrix expediently. The design is proved to be availability by an example about rotating machinery fault diagnosis.