Study On Health Monitoring And Life Prediction Of Power Battery Pack

Since the energy crisis and environmental pollution became more and more serious,the vigorous development of electric vehicles has become an inevitable trend.As the energy source of electric vehicles,power batteries have become the focus of research.Research on the status prediction of battery based on intelligent algorithms has attracted extensive attention from scholars at home and abroad.In this thesis,on the basis of the methods of neural network algorithm,joint method,a method is proposed for predicting state of charge(SOC),state of health(SOH)and remaining useful life(RUL)of batteries..By means of experiment and simulation,many difficult problems are deeply studied like low accuracy,limitation of single parameter prediction for battery health status and remaining service life.The main research work and innovations of this thesis are as follows:(1)The measurement experiments show the internal resistance of lithium batteries has animportant influence on the prediction of SOC which is seldom studied in the existing documentation.In this thesis,BP neural network method is used to set up the prediction model,with the voltage,current,internal resistance and temperature of the acquisition battery as the input of the network while taking battery SOC as the output of the neural network.Compared to the former studies,the prediction model with internal resistance as input in this article has higher accuracy and smaller error fluctuation than the model without internal resistance.(2)On the basis of research work(1),estimated SOH of battery,the improved capacity method,the improved internal resistance method and the voltage method are proposed.Compared to the existing forecasting methods,the prediction time of these three methods has been greatly shortened.The simulation result verifies the higher predictionaccuracy of these three methods.Then,a combined method to estimate the SOH of battery is raised to validate the superiority of the combined method of genetic neural network to the combined method of BP neural network and particle swarm optimization neural network in predicting SOH.(3)On the basis of research work(1)and(2),the remaining useful life battery,curve fitting,Kalman filter and grey neural network have been proposed respectively.After that,a combined method based on neural network is proposed.The simulation results confirm the higher prediction accuracy of the combined method based on neural network than the other three.At the same time,simulation comparison demonstrates the advantages of particle swarm optimization neural network combined method compared with BP neural network combined method and genetic neural network combined method in predicting RUL.