Research On Fault Diagnosis And Health Monitoring Of Power Battery Charging System

With the popularity of electric vehicles,the support of national policies and the inflow of market funds,the electric vehicle charging station and its operation management system have been greatly developed,and the electric vehicle charging fire events occur frequently,and the charging safety of electric vehicles has been widely concerned.Because of the characteristics of the battery management system of electric vehicle,such as large performance difference,off-line delivery and difficult to control,its BMS is not reliable in charge safety assessment.In order to evaluate the health status of power battery effectively,advanced information processing technology and safety evaluation algorithm are needed.Therefore,this paper takes the message information of electric vehicle charging as the research object,first analyzes the real-time status of the message,and then deeply studies the battery health status and charging safety status evaluationFirstly,a fault identification method based on fuzzy neural network and sequential evidence theory is proposed for the data of off line in the charging message of electric vehicle.After statistical analysis and feature extraction of dropped messages,FNN is used to calculate the fault prediction probability of each charging point,electric vehicle and operator,and membership function is used to calculate the corresponding fault membership and uncertainty.Then,DS evidence sequence fusion method is used to realize fault membership fusion,and the corresponding decision criteria are given to get the cause of dropped messages,which is the last step Provide accurate reference information for the operation and maintenance of the project.Secondly,a new SOH estimation method based on the fusion of empirical degradation and radial basis width neural network is proposed to solve the problem of the difference and nonlinear degradation of the healthy state of power cells.The difference of battery health is so great that the empirical degradation model can not accurately estimate the health status of power battery.On the basis of empirical degradation model,combined with RBF＿BLS neural network is used for error compensation to make up for the deficiency of empirical degradation model.The error compensation model fully considers the message data of real-time charging,improves the accuracy of SOH estimation in the process of battery degradation,and provides more accurate information for the follow-up real-time charging safety assessment.Finally,according to different factors affecting the charging safety,a new method of real-time charging risk assessment of electric vehicles based on Improved BP analytic hierarchy process is proposed.Taking the real-time charging message data of electric vehicle as the research object,the membership degree of charging safety influencing factors is analyzed and extracted.The system of charging safety influencing factors is constructed by analytic hierarchy process,and the membership degree of each factor is fused and evaluated.The improved width learning based on compression factor is combined with BP neural network to reduce the training time of neural network and improve the prediction accuracy of neural network.The improved width BP neural network is used to optimize the AHP model to make the charging safety assessment more accurate.