Battery grouping based on improved K-means with curve fitting

摘要

Battery grouping is vital to the performance of the whole battery pack. In this paper, we use curve fitting and improved K-means clustering algorithm for battery grouping. The proposed method uses a set of discharge curves to be clustered. First, we extract several kinds of features of the discharge curves based on the curve fitting and then compute the similarities between batteries according to the features. Finally, a fast search and find of density peaks is utilized to make a pretreatment to feature vector and we use K-means to accomplish the battery grouping. Experimental results show that the proposed battery grouping method is effective.