Echelon utilization of retired batteries based on MMD-based charging curve features difference and state of health

This study introduces a novel framework for the echelon utilization of retired lithium-ion batteries (LIBs) by leveraging the maximum mean discrepancy (MMD) metric to quantify differences in charging curve characteristics. Traditional aging indicators often fail to capture nuanced electrochemical behaviors such as polarization effects. To address this, A MMD metric is proposed to describe the charging curve disparities, derived from an equivalent circuit model and experimental data. Additionally, a genetic algorithm-enhanced K-means (GA-Kmeans) clustering method is integrated to regroup retired LIBs, ensuring high consistency in both aging states and charging behaviors. Experimental results on 32 retired cells demonstrate that incorporating MMD reduces capacity variance within regrouped packs by 54.5% (from 0.0022 to 0.0010) and charging curve inconsistency by 99% (from 0.0939 to 0.0009) compared to conventional methods.