MSFF-DM-based SOH Estimation for Lithium-ion Battery Packs Utilizing System and Cell Features

Accurate estimation of the state of health (SOH) is crucial for safe operation and lifespan extension of lithium-ion battery packs. A multiscale feature fusion decision model (MSFF-DM)-based approach for SOH estimation of lithium-ion battery packs utilizing both system-level and cell-level features is proposed. First, the current, voltage, and temperature data of the lithium-ion battery pack are measured during the charging stage, and the interquartile range (IQR) is extracted through incremental energy (IE) analysis to characterize the pack-level system characteristics. Based on the individual cell temperature and voltage, a combined temperature-weighted voltage inconsistency is constructed to capture inter-cell differences. Subsequently, an MSFF-DM estimation model composed of a multiscale feature extraction layer, feature fusion layer, and decision-making layer is developed. After the two types of features are input into the model for training, the experimental data are used for testing to achieve accurate SOH estimation for the battery pack. The results showed that the average absolute error and root mean square error of SOH estimation using the proposed method are 0.36% and 0.44%, respectively, and the average coefficient of determination exceeded 99%. These results demonstrate that the proposed method outperforms the traditional approaches.