Early detection of internal short circuit faults in lithium-ion battery packs using dynamic time warping and Gaussian mixture clustering

Abstract

Internal short circuit (ISC) faults are a primary trigger for thermal runaway in energy storage lithium-ion battery systems. Timely detection of ISC faults at their early stage can effectively prevent severe safety incidents, thereby ensuring the safe and stable operation of battery energy storage systems. To address this challenge, this study proposes a novel early ISC identification method for lithium-ion battery packs based on dynamic time warping (DTW) sequences and Gaussian mixture model (GMM) clustering. First, the median terminal voltage curve is derived from sorted terminal voltage measurements, serving as a reference representing the normal state of cells within the battery pack. Subsequently, sliding time windows are applied to extract subsequences of the median terminal voltage and individual cell terminal voltages. On this basis, the DTW sequences of each cell are computed and utilized as an indicator to characterize abnormal battery behavior, thereby amplifying the discrepancy between early ISC-affected cells and normal ones. Furthermore, an automatic early ISC fault detection model is developed using a GMM-based clustering algorithm to distinguish between normal and early ISC cells within the battery pack. Experimental validation and analysis under various early ISC fault scenarios with different severity levels demonstrate that the proposed method achieves accurate identification of early ISC cells when the short-circuit resistance is less than or equal to1000 Ω.