Adaptive Neural Fuzzy Petri Net Algorithm for Motor Fault Diagnosis

Abstract

This study aimed to improve the application of fuzzy Petri net to fault diagnosis of motor systems. An adaptive Neural Fuzzy Petri Network Algorithm based on the traditional Petri net theory, fuzzy theory, and neural network algorithm is proposed and applied to the diagnosis of motor faults. The transition confidence is replaced by a Gaussian function to solve the uncertainty of fault propagation. Combined with the BP neural network, fault diagnosis parameters are adaptively trained. Finally, the Neural Fuzzy Petri Net Algorithm is applied to the fault diagnosis of a three-phase asynchronous motor, considering its fault operation mechanism and fault characteristics. The results show that the algorithm can diagnose the fault of the three-phase asynchronous motor with satisfactory accuracy and adaptability.