A sorting method of retired lithium-ion batteries using the improved k-means algorithm based on the incremental capacity curve

Lithium-iron phosphate batteries retired from electric vehicles still have 80% nominal capacity and can be used in utilization. To ensure the electrical performance in second-use utilization, the batteries need to be sorted and regrouped. Effective methods often take a long time and increase the cost. The incremental capacity (IC) curve is derived by calculating the derivative of the charging capacity to battery voltage, which is widely used to analyze battery aging mechanisms. This study found that the similarity of “peaks” in the IC curve can reflect the consistency of voltage and capacity loss. Therefore, a retired battery sorting method based on the IC curve is proposed and applied to 320 retired 18650 batteries. In this method, the K-Means algorithm is first used for clustering, and the T test was used to further improve the clustering degree within the group. The mean square error of each group was reduced by more than 60%. Finally, the results are verified by experiments, which proves the feasibility of the IC curves in retired battery sorting.