Scholarlink

Retired battery laddering is an effective means to maximize the value of batteries. Targeting poor consistency of retired batteries, combining the comprehensive characteristics of discharge depth, charging capacity, open-circuit voltage and charge/discharge curves, a multi-stage sorting method for retired batteries is proposed. In the first stage, outliers are removed using the DBSCAN algorithm; In the second stage, the depth of discharge, open-circuit voltage and charging capacity are used as static features, and the K-means algorithm is used for clustering to realize the coarse sorting of retired battery; In the third stage, the ohmic internal resistance of different SOCs under HPPC conditions is used as a dynamic feature, and the t-Distributed Stochastic Neighbor Embedding(t-SNE) is used for dimensionality reduction, and the category labels of retired batteries are obtained through clustering computation to realize the segmentation of retired batteries for selection. Finally, experiments were conducted using 60 retired batteries, and the consistency of the sorted retired batteries was significantly improved.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-stage Feature Clustering Approach for Sorting Retired Batteries

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Zhuo Liu,
  • Bumin Meng,
  • Xianguang Luo,
  • Xiangyu Xiao,
  • Xuelian Wang,
  • Kaiyu Luo,
  • Da Zhang,
  • Rui Pan

摘要

Retired battery laddering is an effective means to maximize the value of batteries. Targeting poor consistency of retired batteries, combining the comprehensive characteristics of discharge depth, charging capacity, open-circuit voltage and charge/discharge curves, a multi-stage sorting method for retired batteries is proposed. In the first stage, outliers are removed using the DBSCAN algorithm; In the second stage, the depth of discharge, open-circuit voltage and charging capacity are used as static features, and the K-means algorithm is used for clustering to realize the coarse sorting of retired battery; In the third stage, the ohmic internal resistance of different SOCs under HPPC conditions is used as a dynamic feature, and the t-Distributed Stochastic Neighbor Embedding(t-SNE) is used for dimensionality reduction, and the category labels of retired batteries are obtained through clustering computation to realize the segmentation of retired batteries for selection. Finally, experiments were conducted using 60 retired batteries, and the consistency of the sorted retired batteries was significantly improved.