The impact of power battery inconsistency is a significant factor contributing to battery system failure.This inconsistency severely hampers the performance of power batteries and the safe operation of electric vehicles.To address this issue,we propose a method for analyzing the degree of inconsistent faults in power battery cells.Our method leverages ordering points to identify the clustering structure(OPTICS)clustering and voltage anomaly index,using data sourced from automobile enterprise monitoring platforms.We begin by calculating the discrete Fréchet distance and voltage deviation as characteristic parameters for fault diagnosis.Following this,we employ the OPTICS algorithm to cluster these fault features.Using statistical methods,we determine a fault threshold that enables us to accurately identify potential anomalous monomers.The final step involves calculating the voltage anomaly index of the abnormal cell to quantitatively evaluate the battery's fault degree.Our results demonstrate that this diagnosis method does not generate false alarms for normal vehicles,while it accurately identifies all potential abnormal units in vehicles flagged with a"poor battery unit consistency"alarm.When compared with the monitoring platforms of automobile enterprises,our method can detect faults nearly 7 d in advance at the earliest.Moreover,the reliability of our diagnostic results surpasses those of the K-Means++model and the model describing density-based spatial clustering of applications with noise(DBSCAN)even when they use a dynamic K value.This suggests that our method holds substantial promise for application in engineering.
power batterycell inconsistencyfault degree analysisOPTICSvoltage anomaly index
维普
万方数据
