Remaining Useful Life Prediction of Lithium-Ion Batteries Using Support Vector Regression Optimized by Artificial Bee Colony

The remaining useful life (RUL) of LIBs is important in the prognostics and health management of battery systems. However, an accurate RUL prediction is difficult to achieve. Using experimental historical data, this article builds a battery degradation model for estimating battery working state and maintaining and replacing equipment in a timely manner to ensure a stable operation. A method for predicting the RUL of LIBs that employs artificial bee colony (ABC) and support vector regression (SVR) is proposed to improve prediction accuracy. SVR can deal with problems such as small samples, nonlinearity, and time-series analysis. However, SVR is problematic when applied to kernel parameter selection. The ABC algorithm is accordingly employed to optimize the SVR kernel parameters. A simulation with experimental data is conducted by utilizing the NASA Ames Prognostics Center of Excellence battery datasets to validate the proposed method. Results show that parameter optimization with the ABC algorithm is better than that with the PSO algorithm. Furthermore, the ABC-SVR method is more accurate than PSO-SVR and other existing methods are. Therefore, the proposed method achieves high prediction accuracy and prediction stability when used to predict the RUL of LIBs.