Electric vehicle battery pack micro-short circuit fault diagnosis based on charging voltage ranking evolution

Abstract

Micro short circuit (MSC) in Li-ion batteries is characterized by slow development, and usually, MSC fault does not cause significant voltage fluctuations in the early stage. Therefore, early and accurate identification of MSC-faulty batteries is difficult. This paper proposes an MSC fault diagnosis method based on the evolution of the battery charging voltage ranking within multiple charging sections. The ageing trajectory of parameters with significant contributions to the battery terminal voltage is analyzed and proposes the hypothesis of constant battery charging voltage ranking. The battery data are pre-processed using wavelet denoising, and the 3-σ criterion performs the first layer of voltage anomaly diagnosis. The cell charging voltage is ranked horizontally at each sampling point to obtain a median ranking value representing the cell's voltage ranking level in that charging section. The voltage ranking variation factor is constructed, and the 3-σ criterion is used again to detect fault cells with abnormal voltage ranking variation. The method's effectiveness is verified using the actual faulty vehicle data collected. The results show that the method can accurately identify MSC faulty cells despite the absence of voltage anomalies and apparent inconsistencies in the faulty cells.