Research On Life Prediction Of Lithium-ion Battery In Electric Vehicles Based On Gray-AR Model

As the power source of electric vehicles(EV),the lithium-ion battery's performance has a significant effect on EV.It is an important basis for battery management system reliability to predict the lithium-ion battery cycle life,and it is of great significance for the maintaining to extending battery life.According to the fact that the amount of degradation data sample is small,and the degradation path is non-linear and random,a gray-AR(Auto Regression,AR)model for lithium-ion battery life prediction is proposed.The main research work can be divided into three parts.(1)This paper introduces the structure and working principle of the lithium-ion battery,and analyses the degradation mechanism of cells deeply,then explains the factors that have great influence on the degradation.Since the most intuitive change caused by battery performance degradation is capacity degradation,we set the capacity degradation as the feature to characterize the battery life fading.(2)This paper studies the gray system theory and time series analysis theory,and sets both theories as the methods of life prediction for lithium-ion battery in EV.The gray GM(1,1)model and the AR model are eatablished respectively to forecast the battery life using cycle life degradation data of lithium-ion battery from the NASA PCoE.By comparing the prediction result with the actual battery fading data,we can verify the validity of the models.(3)According to the characters of lithium-ion battery capacity degradation path,we use the gray-AR model to predict the battery life.Firstly the GM(1,1)gray model is used to forecast the trend term and the AR model is used to forecast the stochastic term.Then,we combine the two models to form the gray-AR prediction model.Lastly,the prediction result of the gray-AR model is compared with the single model,and the mean absolute error(MAE)and root mean square error(RMSE)are introduced to evaluate the results of this method.The results show that the performance of gray-AR model is better than the two single models,and the prediction accuracy is higher.In order to further prove the advantage of this method,it is compared with the standard PF algorithm.The results shows that the gray-AR model has a higher prediction accuracy than the standard PF algorithm.