Fast Sorting Method of Retired Lithium-Ion Batteries Based on Electrochemical Impedance Spectroscopy

Efficiently sorting large-scale retired batteries is essential for their echelon utilization. Speed and accuracy are critical factors that influence the economic viability, recycling potential, and safety of this process. Current methods, which involve extensive testing and multiple procedures, tend to be overly complex and time-consuming. Thus, enhancing cost-effectiveness in the sorting process while maintaining accuracy presents a significant challenge. This paper introduces a method based on Electrochemical Impedance Spectroscopy (EIS), where features that characterize the consistency of retired batteries are extracted from EIS data. A multilayer perceptron (MLP) model is employed to estimate both SOH and direct current internal resistance (DCIR), which are subsequently utilized in the Autoencoder-FINCH (AEFINCH) clustering algorithm for final sorting. The effectiveness of the proposed method has been validated using a self-tested dataset comprising 381 retired lithiumion batteries. In comparison to state-of-the-art sorting methods, the proposed method achieves reductions of up to 22.06% in SOH standard deviation and 16.74% in DCIR standard deviation.