Machine Learning Models for Accurate Remaining Useful Life Prediction of Lead-Acid Batteries Using Feature Selection

Accurate prediction of the Remaining Useful Life (RUL) of lead-acid batteries is critical for enhancing the reliability and efficiency of electric vehicles (EVs) and energy storage systems. This research focuses on machine learning (ML) models to improve the accuracy of RUL estimation by leveraging feature engineering and model ensemble techniques. A comparative analysis of various ML algorithms, including Random Forest with metric variables, using battery characteristics or indicators such as voltage, current, Charge-Discharge Cycles, and Temperature Variations to track battery health. applying time-series analysis and feature selection to validate the best model that gives optimal and best prediction based on the dataset. The experimental results demonstrate that optimized ML models significantly outperform traditional methods, providing high-precision RUL estimations with reduced error margins. The findings contribute to advancing predictive maintenance strategies, minimizing unexpected battery failures, and extending the operational life of lead-acid batteries in EV applications.