Fault Detection and Identification of Lithium-Ion Battery Based on Improved Multiscale Fuzzy Distribution Entropy

Abstract Short circuits (SCs) in lithium-ion batteries (LIBs) can result in performance degradation, overheating, and even catastrophic events such as fires or explosions. Therefore, quick warning of SCs and accurate fault type identification are essential to ensure the safe operation of electric vehicles. This article proposes a fault diagnosis method based on improved multiscale fuzzy distribution entropy (IMFDE), combined with improved alpha evolution (IAE) optimization algorithm and random forest (RF) to quickly detect battery SCs and accurately identify their types. First, different types of SCs are simulated to obtain fault voltage data. Second, the voltage data are selected by a time window, and IMFDE is extracted as the fault feature. Fault detection is performed by calculating the entropy value of the faulty cell with a deviation metric from the feature of all individual cells in the battery pack. Finally, the RF model is optimized via the IAE algorithm to improve the identification accuracy of battery short circuit type. The proposed method is validated using a large amount of experimental data. The results demonstrate that the proposed method can realize the detection and type recognition of SCs quickly and accurately.