Research of Self-Learning Petri Nets Model for Fault Diagnosis Based on Rule Generation

Depending on the diagnostic rules derived from the default rule generation method of Skowron, a technique to establish Petri net model for fault diagnosis is researched in this paper. In order to simplify the Petri nets model, rule generation need the reduced sample set. However, the reduction of the sample set may cause some errors because of the incompletion of the set. The method can resolve the problem and empower the model the ability of self-learning. The model can auto-update the structure and incidence matrix of the Petri net when diagnostic rules are changed. The method is proved to be available by an example about rotating machinery fault diagnosis in the paper.