Lithium-ion batteries have become the cornerstone of modern energy storage systems due to their high energy density, long cycle life, and efficiency. However, their performance deteriorates over time due to various operational and environmental stresses, leading to unexpected failures and reduced reliability. Accurate Remaining Useful Life (RUL) prediction is critical to ensure the safety, performance, and cost-effectiveness of such battery systems, particularly in safety-critical applications such as electric vehicles, renewable energy storage, and smart grids. This project presents an online fault-tolerant RUL prediction strategy for lithium-ion batteries using advanced machine learning techniques. The proposed system integrates data preprocessing, model selection, and threshold-based health classification to deliver reliable and interpretable predictions. A user-friendly web application enables real-time prediction by capturing essential battery parameters, including discharge time, voltage variations, charging time, and cycle index. Multiple machine learning models such as Random Forest, Extra Tree Regressor, XGBoost, and Gradient Boosting were evaluated, with the Extra Tree Regressor demonstrating superior performance, achieving near-perfect R² and minimal mean squared error. The system classifies battery health into three categories—Healthy, Warning, and Critical—based on the predicted RUL, enabling proactive maintenance decisions. Furthermore, its fault-tolerant architecture ensures robust operation even under uncertain or noisy data conditions. This research contributes to the development of intelligent battery management systems by improving prediction accuracy, enhancing operational safety, and reducing downtime through timely maintenance interventions.

Keywords

Lithium-ion batteries, Remaining Useful Life (RUL), Machine Learning, Fault Tolerance, Predictive Maintenance, Extra Tree Regressor, Online Prediction, Battery Health Monitoring, Smart Energy Systems, Prognostics.