AI-Based State-of-Charge and State-of-Health Estimation for Lithium-Ion Batteries

Abstract

Accurate estimation of battery State-ofCharge (SOC) and State-of-Health (SOH) is essential for the safe and efficient operation of lithium-ion batteries used in electric vehicles, renewable-energy storage, and portable systems. Direct measurement of these internal states is difficult during normal operation because battery behaviour changes with temperature, load, ageing, and operating history. Traditional approaches such as coulomb counting, open-circuit-voltage methods, equivalent-circuit models, and Kalman-filter techniques can provide useful estimates, but their performance may depend on model parameters, sensor quality, and operating conditions. This project presents an Artificial Intelligence-based framework for estimating SOC and SOH from measurable battery signals such as voltage, current, temperature, cycle information, and selected derived features. The proposed system performs data preprocessing, feature extraction, model training, validation, and real-time inference using machine learning and deep learning techniques. Sequenceaware models such as LSTM and GRU can learn temporal relationships in battery measurements, while ensemble models can provide strong performance on engineered features. The system evaluates estimation quality using metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R². By learning nonlinear relationships from historical battery data, the proposed framework supports more responsive battery monitoring, early degradation awareness, improved charging decisions, and intelligent Battery Management System (BMS) operation. The approach provides a scalable foundation for AI-assisted battery diagnostics and energymanagement applications.