A DTW-OPTICS based Multi Fault Diagnosis Method for Battery Packs

This paper investigates the multi-fault diagnosis problem of lithium-ion battery pack. Dynamic time warping and mean voltage deviation are introduced to extract discrepancy features between individual battery mean signal and the average battery voltage signal. The introduction of the average voltage curve effectively reduces computational complexity. Then, an improved ordering points to identify the clustering structure (OPTICS) algorithm is proposed for fault recognition, in which the density of feature points is calculated to determine the threshold for reachability distance, rather than relying on empir-ical settings. Finally, experiments validate the feasibility of the proposed fault diagnosis method.