Research on Calculation Method and SHAP Value Analysis of Lithium Battery Cycle Aging under Irregular Operating Conditions Based on Deep Learning

In the modeling of battery capacity degradation, simplified and fixed testing conditions are often used. Although this approach makes the originally complex calculation problem of battery capacity degradation easier to handle, it may also bring errors due to the inability to accurately reflect the actual situation. This article addresses this issue by providing a more accurate modeling of the battery cycle aging process, taking into account various key factors that affect battery capacity degradation, and without simplifying the discharge and charging process in a single cycle. Then, a deep learning-based method for calculating battery cycle aging was proposed, and feature analysis based on SHAP values was used to enhance interpretability during the training process of the neural network. The proposed algorithm can more accurately and real-time calculate the single cycle aging amount of batteries under current actual operating conditions.