Estimating Lithium Ion Battery Health Metrics: A Data-Driven Approach for SoC, SoH, and RUL Estimation

This work examines the optimisation of forecasting the state of charge (SoC), state of health (SoH), and remaining useful life (RUL) of lithium-ion batteries by advanced machine learning approaches to enhance battery performance, lifespan, and safety in diverse applications. This paper proposes a machine learning approach utilising the Hawaii Natural Energy Institute (HNEI) dataset, employing filter based method for feature selection, voltage and current measurements for synthetic feature generation, and Bayesian optimisation for hyperparameter tuning to predict State of Charge, State of Health, and Remaining Useful Life. The dataset comprised diverse battery cycle indicators that were pre-processed to improve forecast accuracy. Results indicate that the optimised model produced Remaining Useful Life forecasts, SoC and SoH with a root mean square error (RMSE) of 5.5329, 0.0246 and 28.0671 which suggests that the model is able to estimate the parameters correctly. The findings indicate that integrating hyper parameter methods with feature engineering techniques effectively enhances battery health management. This results in improved maintenance strategies and more precise estimations of battery longevity. This study establishes a foundation for future developments in battery management systems and their utilisation in electric cars and renewable energy systems.