Hybrid CNN - LSTM Networks for State of Health (Soh) Estimation of Lithium-Ion Batteries

作者：N Nwankwo Prince, C Okezie Christiana, A. N. Isizoh, Akpado K.A., Alumona T. L · 发表于：International Journal of Advances in Engineering and Management · 年份：2026 · DOI：10.5281/zenodo.22131931 · 研究领域：Advanced Battery Technologies Research、Machine Fault Diagnosis Techniques、Advancements in Battery Materials

Reliable State of Health (SOH) estimation is essential for enhancing the safety, operational reliability, and service life of lithium-ion batteries within Battery Management Systems (BMS). This study proposes a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) model for SOH estimation using the CALCE-2 Battery Degradation Dataset. The battery data were preprocessed through cleaning, Min–Max normalisation, feature engineering, and sequential windowing before model training. The proposed CNN–LSTM architecture makes good use of the feature extraction ability of CNNs and the temporal learning ability of LSTMs to capture complex battery degradation patterns from voltage, current, charge capacity, discharge capacity, internal resistance, and impedance measurements. The model was evaluated using the Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Coefficient of Determination (R²). The proposed model achieved an MAE of 0.020259, RMSE of 0.024045, and R2 of 0.875912, indicating an accurate SOH estimation and strong agreement between the predicted and reference values. The proposed framework offers a robust approach for intelligent battery health monitoring and predictive maintenance in Battery Management Systems (BMS).