Research On Intelligent Fault Diagnosis Strategy Of Electric Vehicle Charging Facilities

In 2022,the sales of new energy electric vehicles in the country grew faster than the previous year.Driven by national and local policies,the development momentum of electric vehicles will continue to rise,and it is expected that the sales of electric vehicles will further increase in 2023.The safety of charging facilities is becoming more and more important.If charging facilities fail frequently,the safety of EV users and the further development of EV are bound to be seriously affected.Therefore,it is of great significance to study the fault diagnosis of electric vehicle charging facilities.In this paper,the intelligent fault diagnosis strategy of charging facilities is studied.Firstly,it introduces the development background of charging facilities,the research status of intelligent diagnosis strategy and fault diagnosis of charging facilities at home and abroad,and describes the chapter arrangement of this paper.Secondly,it expounds the working principle of charging facilities,analyzes the faults and defects in the charging process of charging facilities and their causes,and establishes FMEF fault analysis table.Then,two kinds of intelligent diagnosis strategies are proposed for fault diagnosis of charging facilities,one is inference model based on rule engine,the other is intelligent diagnosis based on GRU neural network.Then,considering the advantages of Drools rule engine,such as declarative programming,scalability and high diagnostic efficiency,reasoning model is built based on Drools rule engine,charging facilities and EV charging related standard data are consulted,and fault rule base of charging facilities is designed for reasoning of rule engine.Using "triple" data representation for data transmission and programming,improve the processing efficiency of inference engine.Then,considering the limitations of the rule model,in order to make full use of the characteristics of the charging data stored by the charging facility,the GRU neural network model suitable for time series is adopted for diagnosis,the Adam algorithm is used to optimize the network iteration process,and the Dropout method is used to prevent overfitting of the training model.Finally,considering the inadequacies of neural network random initial parameters,which affect the fitness of model training,the network parameters are optimized by combining the advantages of genetic algorithm and ant colony optimization algorithm.Based on the characteristics of selection,crossover and mutation of genetic algorithm,the limitations of local optimal parameters are jumped out to improve the diagnostic performance of the training model.Based on the ant colony optimization algorithm "pheromone" constantly updated characteristics,improve the network model parameters,and the model diagnosis effect before and after optimization,the results show that the genetic algorithm combined with ant colony algorithm optimization GRU network can effectively improve the accuracy of fault diagnosis.