A Novel Multifault Diagnosis Method for Lithium-Ion Batteries Based on Cloud Low-Density Segment Charging Voltage Curves

Timely and accurate multi-fault diagnosis of lithium-ion battery systems is crucial for ensuring safety in large-scale applications. This paper presents a new multi-fault diagnosis method based on cloud low-density segment charging voltage curves. Firstly, the sampling drift fault of the voltage sensor is detected and repaired based on the voltage sampling principle in the series-connected battery module. The segment charging voltage curves are extracted, and then the fused feature is constructed using mean normalization and local outlier factor feature of the segment charging voltage curves. The multi-fault of low-capacity, low state of charge, and micro short-circuit faults in the series-connected battery pack are diagnosed with the adaptive dynamic threshold and 3σ criterion, the detection accuracy can be up to 98.6%, 98.7%, and 98.2%. This method based on partial charging voltage curve does not rely on charging conditions and completed charging curves, ensuring that the batteries can be frequently monitored during daily use to improve system safety.