汽车工程 ›› 2024, Vol. 46 ›› Issue (4): 643-651.doi: 10.19562/j.chinasae.qcgc.2024.04.010

• • 上一篇    

基于自适应模糊C-均值算法的退役锂离子电池快速聚类

陈琳1,2,何熳平1,吴淑孝1,陈德乾1,赵铭思1,潘海鸿1()   

  1. 1.广西大学机械工程学院,南宁 530004
    2.广西制造系统与先进制造技术重点实验室,南宁 530004
  • 收稿日期:2023-07-20 修回日期:2023-08-30 出版日期:2024-04-25 发布日期:2024-04-24
  • 通讯作者: 潘海鸿 E-mail:hustphh@163. com
  • 基金资助:
    国家自然科学基金(52067003);广西制造系统与先进制造技术重点实验室项目(22-035-4S013)

Fast Clustering of Retired Lithium-ion Batteries Based on Adaptive Fuzzy C-means Algorithm

Lin Chen1,2,Manping He1,Shuxiao Wu1,Deqian Chen1,Mingsi Zhao1,Haihong Pan1()   

  1. 1.School of Mechanical Engineering, Guangxi University, Nanning 530004
    2.Guangxi Key Lab of Manufacturing System and Advanced Manufacturing Technology, Nanning 530004
  • Received:2023-07-20 Revised:2023-08-30 Online:2024-04-25 Published:2024-04-24
  • Contact: Haihong Pan E-mail:hustphh@163. com

摘要:

梯次利用处理退役锂离子电池具有巨大的经济和环境价值,而如何高效、准确地对退役电池进行分选重组是梯次利用中突出的技术挑战。首先,为准确反映退役电池的一致性,提取最大可用容量(MAC)、放电欧姆内阻(DOIR)和容量增量曲线的弗雷歇距离(FD)3个因素共同作为聚类因子。然后3个聚类因子结合自适应模糊C-均值(AFCM)算法构建退役电池聚类方法。结果表明:AFCM算法聚类簇内MAC的最大误差为79 mA·h,DOIR小于45 mΩ;三因素的聚类方法成组的电池一致性较好;并且在117颗电池聚类时,AFCM算法聚类耗费的时间最短。

关键词: 退役锂电池, 梯次利用, 重组聚类, 机器学习

Abstract:

The treatment of retired lithium-ion batteries (LiBs) by echelon utilization has great economic and environmental values, and how to sort and reconstitute decommissioned batteries retired LiBs efficiently and accurately is a prominent technical challenge in stepwise utilization. Firstly, to accurately reflect the consistency of retired batteries LiBs, the three factors of maximum available capacity (MAC), discharge ohmic internal resistance (DOIR) and Frechet distance (FD) of incremental capacity curve, are extracted together as clustering factors. Then the three clustering factors are combined with the adaptive fuzzy C-mean (AFCM) algorithm to construct a clustering method for retired batteries LiBs. The results show that the maximum error of MAC within the clustered clusters of the AFCM algorithm is 79 mAh with the DOIR less than 45 mΩ. The clustering method of the three factors into groups of batteries has better consistency; and the AFCM algorithm clustering takes the shortest time when 117 batteries are clustered.

Key words: decommissioned lithium batteries, echelon utilization, sorting and restructuring, machine learning