Fault Diagnosis Of Electric Vehicles Based On GrC-Neural Network And Evidence Theory

In the increasingly severe context of the global environmental pollution and fossil energy crisis,the electric vehicle becomes the new direction of automobile industry all around the world.The safety and reliability of electric vehicles are important indicators to measure their quality.Making judgments on the running status and abnormality of electric vehicles,fault diagnosis technology enhances the safety performance of electric vehicles and ensure driving safety.Therefore,further study of electric vehicle fault diagnosis technology has important theoretical significance and application value.Faced with the problem of complexity and non-linearity of the fault data of the electric vehicle,in this dissertation,an electric vehicle fault diagnosis method based on granular computing-neural network(GrC-NN)and DS evidence theory is proposed to simplify the neural network structure and improve the accuracy of electric vehicle fault diagnosis.The main works are as follows:(1)Research on the fault diagnosis methods of electric vehicles.The causes and common faults of electric vehicle faults are qualitative analyzed,and electric vehicle faults are classified;the common fault diagnosis method for electric vehicles are investigated.(2)The fault diagnosis method of electric vehicle based on particle calculation.The knowledge reduction algorithm based on granular matrix is analyzed;the principles of GrC-BP neural network and GrC-RBF neural network are introduced respectively;the process of fault diagnosis method for electric vehicles based on GrC-NN is analyzed;it is proved by simulation that the particle calculation can improve the training speed of neural network and simplify the structure of neural network model while basically keeping the diagnostic accuracy.(3)The fault diagnosis method of electric vehicle based on GrC-NN and DS evidence theory.The problems of evidence paradox and improved evidence synthesis rules for evidence fusion are analyzed;the process of electric vehicle faults diagnosis method based on GrC-NN and DS evidence theory is analyzed;compared with the simulation method of electric vehicle fault diagnosis based on GrC-NN,the method in this dissertation has better accuracy and lower diagnostic uncertainty;the effectiveness of the proposed method is further verified by experiments under different load conditions.