Rapid Classification Based on Fast Charging Curves for Reuse of Retired Lithium-ion Battery Modules

Owing to the inconsistent decay among cells during their applications, lithium-ion batteries need to be classified before their cascade utilization in energy storage stations. However, the existing classification methods have a low efficiency and low accuracy. Moreover, the classification researches on the module level are relatively deficient compared with those on the cell level. This paper mainly proposes a rapid classification method based on the machine learning algorithm for retired lithium-ion battery modules. First, the classification procedures are designed based on the characteristics of the series-charging curves of retired cells. Then, the classification model is established with the support vector machine screening to estimate the voltages of a large scale of cells from those of a few sample cells. Moreover, a regrouping approach aiming at the cell modules is introduced to solve the problem of disassembly complexity in practical applications. The results show that, compared with the traditional methods, the classification efficiency are increased significantly with the effect of module level regrouping improved. © 2020, Power System Technology Press. All right reserved.