Lithium ion battery is a key component of energy storage system. Accurate and scientific prediction of its Remaining Useful Life (RUL) is an important factor to check the operation of energy storage system is whether reliable. ARIMA is an effective time series prediction processing method, which can be used to calculate battery RUL and its confidence interval. And the more predicted samples, the higher the prediction accuracy. Compared with the empirical model and support vector machine algorithm, the analysis results show that the support vector machine is over-fitting. For two sets of the experimental data, the absolute predictive error of ARIMA algorithm is approximately 1.2%, that of linear model is approximately 1.4%, and that of Verhulst model is approximately 7.5%, which verifies the accuracy of ARIMA time series model in predicting the RUL in long interval.
Maximum Available Capacity and Energy Estimation Based on Support Vector Machine Regression for Lithium-ion Battery
