Battery state of health estimation using deep learning and data-driven approaches

This study investigates the effectiveness of deep learning and data-driven approaches for the estimation of the State of Health (SOH) of lithium-ion batteries. Accurate SOH estimation is crucial for ensuring the safe and efficient operation of batteries. In this context, various deep learning models, including GRU, LSTM, MLP, CNN, and Attention-based networks, were trained and compared using widely adopted datasets such as NASA and CALCE. Furthermore, a transfer learning approach was employed, where models trained on one dataset were retrained and applied to another dataset. This approach aims to enhance model performance in scenarios with limited data availability. The performance of the models was evaluated using metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and the coefficient of determination (R²). The results demonstrate that deep learning-based methods provide high accuracy in SOH estimation and that retrained models using transfer learning can be effectively applied across different datasets.