An Online Fault Diagnosis Method Based on Multiscale Permutation Entropy for Lithium-ion Battery Pack

Efficient and accurate battery fault diagnosis is the foundation and prerequisite for the safe and stable development of electric vehicles. The vast majority of thermal runaway in electric vehicles is caused by a short circuit fault in the battery, which can lead to fire and explosion. Therefore, it is crucial to detect short circuit fault quickly and accurately. But common fault diagnosis methods are difficult to detect battery abnormality in the early stages. In this paper, an online fault diagnosis method of lithium-ion batteries based on multiscale permutation entropy (MPE) is proposed, which can effectively handle the voltage characterization information at the early stage of fault. The value of MPE is processed to derive the abnormality coefficient (AC), and the minor fault can be diagnosed by comparing the AC of the battery. Compared with the traditional permutation entropy (PE) algorithm, the MPE algorithm can detect the occurrence of faults at multiple scales and can be applied in progressive fault diagnosis. The experiment results indicate that the proposed MPE method can effectively detect short circuit fault in the early stages with high accuracy and strong robustness.