Micro-fault diagnosis of electric vehicle batteries based on the evolution of battery consistency relative position

Abstract

Micro-faults in Li-ion batteries are a safety hazard for battery packs, and accurately identifying micro-faulted batteries is a complex problem to solve. In this paper, we propose a micro-fault diagnosis method based on the evolution of the consistent relative position of cells within multiple charging segments. The CCVC (Charging cell voltage curve) transformation is applied to match the charging voltage curve of each battery with the reference battery, and the particle swarm algorithm obtains the matching transform parameters reflecting the degree of battery consistency with adaptive inertia weights. The transformation parameters are normalized to obtain each battery's relative position Z-Score values in the consistency comparison of the transform parameters. Based on the constant consistency relative position hypothesis, the evolution of the consistency relative position of each cell is scored quantitatively by constructing consistency relative position fluctuation scores. The anomaly detection algorithm based on the 3 − σ criterion is used to identify and locate the micro faulted battery by comparing the scoring results of each battery. The method's effectiveness is verified using the collected actual breakdown vehicle data, and the influence of different reference cells on the technique is analyzed. The results show that the consistency relative position of healthy cells is almost stable over medium-term scales, while the consistency relative position of faulty cells falls. This method can accurately and effectively locate faulty cells even though the battery pack does not exhibit abnormal voltage fluctuations or significant inconsistencies.