Precision Forecasting: Optimizing Lithium-ion Battery Remaining Usable Life Estimation with Cutting-Edge Machine Learning

Abstract

This study seeks to estimate the remaining life span of lithium-ion batteries, which is an essential component of early failure prevention. The paper demonstrates how advanced machine learning tools like CatBoost and LightGBM are superior when it comes to handling complex data patterns. In assessing the accuracy of prediction, several key performance metrics such as Mean Squared Error (MSE), Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) were used. Experimental validation using NASA's 18650 lithium-ion battery datasets reveals a 25% improvement in prediction accuracy, with CatBoost consistently outperforming LightGBM. This implies that these approaches have the potential to improve RUL predictions and thus battery management policies.