The Fault Diagnosis Study And Performance Analysis Of Lithium Power System Battery

With the aggravation and deterioration of energy and environmental problems the whole society is facing,the development and application of electric vehicles are gradually expanding.As the energy source of electric vehicles,power battery system is particularly important in electric vehicles,but it is also the main source of failure of electric vehicles,greatly affecting the normal operation of the vehicle.Therefore,it's important to improve the stability of power battery system and ensure stable output.Firstly,the development and application of electric vehicles and power batteries are introduced,and the research results of fault diagnosis technology are summarized.After analyzing lithium battery principle,parameters,aging cause and fault reasons,a data-driven fault diagnosis method is selected to diagnose the fault of lithium power battery system.The failure level is confirmed by failure mode impact analysis(Failure Mode and Effects Analysis,FMEA),and the necessary measures are determined.The fault of power battery system is very complex and uncertain,and the lack of real-time test points and incomplete test data bring great difficulties to the performance analysis and fault diagnosis of power battery system.In the view of the advantages of support vector machine in small sample data sets,the performance parameters of lithium-ion battery system can be predicted by using PSO-SVM(Particle Swarm Optimization Support Vector Machine,PSO-SVM).With simulation modes,the simulation results show that this algorithm has favorable performance in optimization and prediction,which it's suitable for lithium power battery system working state prediction.Finally,combining with the results of PSO-SVM and FMEA,it's presented to use fuzzy Bayesian network for lithium battery system fault diagnosis,by bring fuzzy theory into na?ve Bayesian.And the simulation results present that the fault reason recognition rate of this algorithm is high,and this algorithm can reduce the conditional probabilities needed in the process of inference,which is fast,feasible and accurate.