An Online Multifault Diagnosis Scheme for Battery Packs Based on Voltage Envelope Relationship

Abstract

Electrical faults pose a serious threat to the safe operation of battery packs. Common electrical faults include undervoltage, overvoltage, connection faults, and sensor faults. However, existing methods fail to provide a comprehensive and adequate diagnosis of the four types of electrical faults due to their inability to distinguish between fault signatures. This article proposes an online multifault diagnosis scheme based on voltage envelopes. First, using the positive and negative envelopes of voltages, unique fault signatures are constructed for each type of electrical fault, where the similarity and polarity of the envelopes are selected as indicators. Then, an improved Hausdorff distance and a cross-voltage measurement circuit are employed to catch the fault signatures. By comparing the similarity and polarity between the adjacent envelopes, four types of faults can be detected and sequentially distinguished. Finally, an online diagnosis flow is designed, where normalization and the Douglas–Peucker algorithm are used to ensure robustness to inconsistency and noise. Experimental results show that the four types of faults can be accurately diagnosed within a short period of time, and a minor fault characterized by a 0.07-C discharge current can still be sensitively detected. In addition, the inconsistency and noise within the normal range cannot cause a misdiagnosis.