A Weighted Feature Fusion-Based SOH Assessment for Lithium-Ion Batteries

The SOH prediction of lithium-ion batteries is crucial for ensuring their safety and extending their lifespan. Traditional SOH prediction methods typically use manual feature extraction combined with regression models, while data-driven deep learning methods have also achieved significant results in recent years. However, there are still challenges in effectively integrating these methods. In this paper, we propose a new approach to lithium-ion battery state of health (SOH) assessment, which combines data-driven features and statistical features with an attention mechanism. By analyzing 128 statistical features and data-driven features, we reveal the importance contributions of features on the assessment of SOH. The average root mean square error (RMSE) of the final group test is 0.98%, and the average mean absolute error (MAE) is 0.66%, indicating that the method can effectively predict the SOH of lithium-ion batteries. The results of this study are helpful to better understand and predict the SOH of lithium-ion batteries, and have important significance for the design and optimization of battery management systems.