Comparative analysis of battery diagnostic methodologies considering parameter learning process

With the continuously growing number of electric vehicles, the safety of electric vehicle battery systems has received wide attention. The threshold parameter learning of the data-based battery diagnosis methods highly affects the algorithm performance but is seldom discussed. In this paper, two commonly used fault diagnosis methods, 3σ multi-level screening method and LOF algorithm are compared to evaluate the fault detection performance, and the learning of the threshold parameters are particularly discussed. The results further address the importance of threshold parameter learning for the performance evaluation of battery diagnosis methods.