LIME-Based Explainable AI Approach for Battery Health Monitoring

It is crucial to monitor battery health to ensure the reliability and longevity of energy storage systems. This study presents an Explainable AI (XAI) powered monitoring framework that combines machine learning with improved interpretability. The NASA battery health dataset was used to train two predictive models, Random Forest and XGBoost, using derived voltage, charge voltage, measured current, charge current, temperature, and battery usage time as the key process indicators. The Random Forest model achieved an accuracy of 96 %, while XGBoost attained 98 %. The method also integrated the Local Interpretable Model-Agnostic Explanations (LIME) framework to add transparency to its decisions, further illustrating the importance of features. Real-life use cases were examined to determine how well the system performs in battery condition monitoring. The proposed AI-powered monitor provides trustworthy predictions with understandable insight into better predictive maintenance and fault detection. The enhanced explainable approach of the proposed work increases battery management systems' utility by addressing crucial aspects of electric vehicles and renewable energy storage systems.