Detection and differentiation of multiple types of minor anomalies in battery packs

Lithium-ion battery packs serve as the primary energy source for electric vehicles and energy storage systems. However, various types of minor anomalies in the battery packs can significantly compromise the safety and stability of these systems. In particular, micro-short-circuit (MSC), low-capacity (LC), and low-electric-quantity (LEQ) are challenging to detect at the initial stage. Moreover, these three minor anomalies exhibit similar voltage changes, which often results in detection confusion. In this paper, a comprehensive detection method for minor anomalies is proposed to achieve accurate detection and differentiation of MSC, LC, and LEQ. First, the Z-score normalization method is employed to enhance the identification of abnormal features within the battery packs. Second, the Z-score curves is smoothed using the Kalman filtering algorithm. Third, a density-based spatial clustering of applications with noise (DBSCAN) algorithm is employed to automatically detect and localize these three types of anomalies. Finally, the calculation and comparison of anomaly duty cycles during the charging stages enable multi-anomaly differentiation. The experimental results demonstrate that the proposed scheme accurately identifies MSC even when the short-circuit degree is minor. Moreover, it can simultaneously detect and differentiate MSC, LC, and LEQ across multiple charging modes.