Remaining-Useful-Life Prediction for Li-Ion Batteries

This paper aims to establish a predictive model for battery lifetime using data analysis. The procedure of establishment is illustrated in detail, including the pre-processing, modeling, and prediction. characteristics lithium-ion batteries are introduced. In this study, analysis performed with MATLAB, open-source provided by NASA. addressed models include decision tree, nonlinear autoregression, recurrent neural network, long short-term memory network. part training, root-mean-square error, integral squared absolute error considered cost functions. Based on defined health indicator, remaining useful life can be predicted. confidence interval used describe level each According test results, network provides best performance among all models.