Periodic Segmentation Transformer-Based Internal Short Circuit Detection Method for Battery Packs

With the rapid evolution of electric vehicles (EVs), assuring the security and dependability of battery packs has acquired paramount significance. Internal short circuit (ISC) within EV battery packs poses a threat to the safety and reliability of EVs. Most of the existing ISC detection methods still suffer from two limitations, i.e., the dataset incompleteness and poor feature representation. To address these challenges, we develop a periodic segmentation Transformer-based ISC detection method for battery packs. First, considering three different operating conditions, a comprehensive dataset encompassing three distinct ISC severity levels is constructed. Second, to facilitate understanding of the proposed model design, a discrete wavelet transform (DWT)-based periodicity analysis module and a time-oriented segmenting module are developed. This dual-module design empowers the model to adjust the length of sliding windows adaptively and enables the joint capture of temporal-spatial and periodic information, significantly enhancing the feature representation ability. Third, the experimental results show that our method outperforms the best state of the art (SOTA) in terms of accuracy (average <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$F1$ </tex-math></inline-formula> score improved by 24.2%). Finally, robustness analysis and generalization analysis are conducted. The former one demonstrates robustness in terms of parameters within the adaptive aggregation module and input data length, and the latter one demonstrates the generality of the feature extraction method.