Rapid Sorting and Regrouping of Retired Lithium-Ion Battery Modules for Echelon Utilization Based on Partial Charging Curves

The lithium-ion battery of an electric vehicle continues to have available capacity even after it is retired, thus representing good echelon utilization value. The ideal regrouping form for echelon utilization is conducted at the module level. However, existing sorting methods are generally only suitable at the cell level. To address this issue, a fast sorting and regrouping method is proposed at the module level based on a machine learning algorithm. First, the correlation between the charging curve and the remaining useful capacity of the battery is investigated. The charging curves of cells in a module are translated and supplemented to extract the capacity characteristics without disassembling the modules. Next, a rapid sorting model based on the support vector machine is proposed to estimate the capacity. Then, a regrouping method based on an improved K-means algorithm that considers different echelon utilization scenarios at the module level is proposed. Finally, simulations and experiments are conducted to verify the effectiveness of the proposed method. The results show that the capacity prediction accuracy is within 3%, and the consistency of the echelon utilization battery system obtained by the proposed regrouping method is higher than that obtained by the conventional method.

Automated summary

A fast sorting and regrouping method based on machine learning algorithm for lithium-ion batteries in electric vehicles is proposed at the module level. By analyzing the charging curves in the module, capacity prediction is achieved, and an improved K-means algorithm is used for regrouping.