Remaining Useful Life Predictor for EV Batteries Using Machine Learning

The swift advancement of electric vehicle (EV) technology enhances the focus on sustainable energy storage and underscores the crucial significance of lithium-ion batteries. This research primarily presents the techniques of forecasting the Remaining Useful Life (RUL) of this battery using advanced Machine Learning (ML) methods such as Random Forest (RF) and Support Vector Machine (SVM). This research centres around the thorough preprocessing of a detailed dataset received from the NASA Ames Prognostics Center of Excellence. The One-way ANOVA method is employed to find the optimum set of features. The exhaustive hyperparameter-tuning (HPT) was performed to boost the performance of the ML models. An important component of this study is its pragmatic methodology, which considered real-time variables such as temperature changes and usage cycles to analyse the effect on battery capacity (cap). The proposed system helped to understand the behaviours of battery deterioration trends more comprehensively. The effectiveness of the system is decided based on the R2 score and Mean Squared Error (MSE). The RF model has shown R2 score of 0.83 and MSE of 1.67. The result enhances lithium-ion battery safety and efficiency by establishing new predictive models. Thus, it provides a better battery management system for electric vehicles. As a result, it promotes the development of more sustainable and economical energy solutions.