An efficient screening method for retired lithium-ion batteries based on support vector machine

Abstract

As a large number of lithium-ion batteries are retired from electric vehicles, their reuse is receiving more and more attention. However, a retired battery pack is not suitable for direct reuse due to the poor consistency of in-pack cells. In this paper, we propose an efficient screening method for retired cells based on support vector machine. Firstly, twelve retired LiFePO4 battery modules are dissembled into 240 cells, and their capacity and resistance are measured and analyzed. Secondly, to improve screening efficiency for retired cells, an incremental capacity curve based on high charging current rate is used to rapidly extract their capacity feature and internal resistance. Subsequently, the multi-class model based on support vector machine is trained to classify the retired cells with good consistency. Finally, the retired cells are accurately divided into four classes by the trained model, and the classification accuracy can reach 96.8%. Compared with the traditional method, the time of feature extraction can be reduced by four fifths, and the screening efficiency is greatly improved. Additionally, a current test system is designed to compare the current differences in the new battery module regrouped in parallel by the screened cells. The experimental results show the current consistency is significantly improved compared to that in the original battery module, and the mean of standard deviation used to describe the current inconsistency drops by up to about 14 times.