Research On Fault Diagnosis Of Power Electronic Circuits Based On Neural Network

The application of electric power electronic installation is day by day widespread,the close and complex degree of its active channel constantly is also expanding,how to look for the appropriate fault diagnosis method,to diagnosis at high speed and highly effectiveness when electric power electronic circuit breaks down,and to reduce breakdown affects as far as possible,became the important topic which could influences electric power electronic industry development.Through various methods which fault diagnosis analysis research use,the paper chose the mainstream using the method based on artificial neural networks to carry on the main object of hypothesis study,introduced the artificial neural networks structure,classification and study mechanism.In most commonly used,the BP network to the neural network has been carried on the special elaboration.The model structure,the movement principle,merit and limitatio n of BP network were analyzed in detail.The paper designed a single-phase bridge rectifying circuit using the Matlab simulation software,constructed the BP neural network to conduct the case study of fault diagnosis to it,and enabled the model to achiev e better diagnosis effect through changing network architecture and parameter.At last,a 3-phase voltage-regulating circuit is chosen as a case to realize the BP neural network based diagnosis scheme.From the simulation result,definitely may use the BP neural network method to carry on the failure diagnosis of electric power electronic circuit.