Hierarchical fault diagnosis of power electronic devices based on fuzzy clustering and artificial neural network

To solve the problems of controllable angle's large change and meet the high real time demand of fault diagnosis in. Three-phase Controlled Rectifier (TCR), a new approach combining clutstering analysis theory and neural networks for fault diagnosis of TCR is presented. This paper introduces the basic theories of fuzzy clustering analysis and neural networks, develops intelligent fault diagnosis technique based on hierarchical clusterting analysis and networks, analyzes the fault types in, TCR, researches on methods which is used to extract and cluster the fault feature information, presents the design of hierarchical fault diagnoses system. based, on clustering and artificial neural networks. The novel method is proved to be valid through testing classic circuit. For the given, fault types of power electronic circuits, the experimental results show that the proposed method, has better fault resolution and can, achieve a high fault diagnosis accuracy of 91.52%.