Fault diagnosis method for lithium-ion batteries in electric vehicles using generalized dimensionless indicator and local outlier factor

Failure of the Lithium-ion (Li-ion) battery pack can affect electric vehicles' safe and reliable operation. Accurate detection of early battery pack failures can prevent safety incidents and reduce property damage. However, detecting early battery pack failures through actual operating data from electric vehicles is still a challenging task. To extract fault information from real EV usage data, a generalized dimensionless indicator (GDI) with a tolerance factor was constructed for this study. Mapping dimensionless indicators into 2-dimensional space for representing patterns of voltage anomaly evolution. The LOF algorithm is used to find the evolutionary pattern of anomalies, identify faulty batteries and determine when the fault occurred. A differential method is used to avoid the effect of cell inconsistencies on diagnostic results. Experimental validation is carried out using real vehicle data. The results show that the proposed method can accurately identify faulty batteries at an early stage of failure and has good robustness.