Battery Life Prediction Based On ARIMA With BPNN

As technology continues to enter our lives,lithium battery has penetrated into various industries.Since the battery explosion of Samsung mobile phone,the safety of lithium battery has always been the focus and hot issue of the society.The prediction of lithium battery life and the improvement of lithium battery performance become the focus of research,so it is necessary to find the most appropriate life prediction method to improve the safety of lithium battery,which is of great significance for the use of lithium battery as energy devices.In the research of lithium battery life,the mainstream research methods at home and abroad are still based on data-driven,and the methods based on degenerate physical model are only suitable for some specific scenarios.In this paper,data-driven method is used to predict the life of lithium battery.The main contents are as follows:Firstly,this paper introduces the research background,describes the internal structure of lithium battery,introduces the working principle of lithium battery,and observes the changes of parameters during the working process.Looking for the influence factors of lithium battery life decline,analyzing the main factors of lithium battery life decline,it is concluded that the battery life decline is affected by many internal parameters.Then,all the influence factors are transformed into the form of battery capacity attenuation.Therefore,the attenuation of battery capacity directly represents the attenuation of battery life.Secondly,the life of lithium battery is predicted by using the autoregressive integrated moving average model(ARIMA),and the existing battery capacity decline data is used as the capacity decline value after the time series prediction.The prediction results show that ARIMA model has strong linear prediction effect,strong short-term prediction ability and prediction error of 4.7%.Thirdly,the back propagation neural network(BPNN)is used to build a model to predict the battery life.The prediction results show that BP neural network can effectively predict the life of lithium battery.The model can learn the historical loss track of the actual capacity of lithium battery,and can use different amount of data as training data to predict the life of lithium battery.The prediction error is 7.03%.Finally,because ARIMA belongs to the type of linear prediction model,the short-term prediction is more accurate and not suitable for long-term prediction;the advantage of BPNN model is that it has strong nonlinear fitting ability,so the combination of the two models will get more accurate results.The original data is processed and predicted by ARIMA to get the prediction result;the error data ispredicted by BPNN to get the prediction result.The final prediction result is obtained by adding the results of the two models.The prediction error of the combined model is3.13%.Compared with the three models,the combined model has the highest prediction accuracy.