储能科学与技术 ›› 2025, Vol. 14 ›› Issue (9): 3541-3551.doi: 10.19799/j.cnki.2095-4239.2025.0221

• 储能测试与评价 • 上一篇    

基于产线大数据的电池内阻预测及快速分选方法

包新宇1(), 孔祥栋2, 吕桃林3, 朱志成1,4, 韩雪冰5, 来鑫1, 郑岳久1, 孙涛1()   

  1. 1.上海理工大学机械学院，上海 200093
    2.上海电机学院机械工程学院，上海 201306
    3.上海空间电源研究所，空间电源国家重点实验室，上海 200245
    4.江西职业技术大学机械工程学院，江西 九江 332007
    5.清华大学车辆与运载学院智能绿色车辆与交通全国重点实验室，北京 100084
  • 收稿日期:2025-03-06 修回日期:2025-04-09 出版日期:2025-09-28 发布日期:2025-09-05
  • 通讯作者: 孙涛 E-mail:1728159144@qq.com;tao_sun531@usst.edu.cn
  • 作者简介:包新宇（2000—），男，硕士研究生，研究方向为锂离子电池产线大数据，E-mail：1728159144@qq.com；

Battery internal resistance prediction and rapid sorting method based on production line big data

Xinyu BAO1(), Xiangdong KONG2, Taolin LV3, Zhicheng ZHU1,4, Xuebing HAN5, Xin LAI1, Yuejiu ZHENG1, Tao SUN1()   

  1. 1.College of Mechanical Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
    2.School of Mechanical Engineering, Shanghai DianJi University, Shanghai 201306, China
    3.State Key Laboratory of Space Power-sources, Shanghai Institute of Space Power-sources, Shanghai 200245, China
    4.School of Mechanical Engineering, Jiangxi Polytechnic University, Jiujiang 332007, Jiangxi, China
    5.School of Vehicle and Mobility, State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University, Beijing 100084, China
  • Received:2025-03-06 Revised:2025-04-09 Online:2025-09-28 Published:2025-09-05
  • Contact: Tao SUN E-mail:1728159144@qq.com;tao_sun531@usst.edu.cn

RichHTML

0

PDF

0

摘要：

基于产线大数据平台，利用容量对出厂电池进行快速分选，保证电池组的一致性，可以提高电池包的使用寿命，但无法保证电池组的动态一致性。本工作提出了一种通过产线大数据来预测分容阶段电池内阻，根据预测出的电池内阻，结合容量对电池进行快速分选的方法。通过收集产线数据，利用皮尔逊相关性分析(PCC)对特征进行筛选，然后利用多层感知机(MLP)构建内阻预测的神经网络模型。将预测出的内阻结合容量数据，利用模糊C均值聚类(FCM)将产线电池分为4个档次，然后利用充放电电压曲线特征作为评价标准，对分选效果进行验证，并与传统的容量分选方法效果进行对比。结果表明，内阻预测的平均绝对百分比误差(MAPE)为1.2%，分选效果的综合优化率达到14.9%，相较于传统容量分选效果有显著提升。本工作为产线电池的快速分选提供了一种新的方法。

关键词: 产线数据, 锂离子电池, 内阻预测, 聚类分析, 电池分选

Abstract:

Based on a production line big data platform, rapid sorting of outgoing batteries using capacity data can improve the consistency of battery packs and extend their service life; however, this method alone does not ensure dynamic consistency. In this study, a novel approach is proposed to predict the internal resistance of batteries during the capacity grading stage using production line big data, followed by rapid sorting that combines predicted internal resistance with capacity data. Specifically, production line data are collected, and relevant features are screened using the Pearson correlation coefficient (PCC). A neural network model based on a multilayer perceptron (MLP) is constructed to predict battery internal resistance. The predicted internal resistance values are then integrated with capacity data, and the Fuzzy C-Means (FCM) clustering algorithm is employed to classify the batteries into four grades. The effectiveness of the proposed sorting method is evaluated using charge-discharge voltage curve characteristics as assessment criteria and compared with the traditional capacity-based sorting method. The results demonstrate that the mean absolute percentage error (MAPE) of internal resistance prediction is 1.2%, and the overall optimization rate of the sorting process reaches 14.9%, representing a significant improvement over traditional capacity sorting. This study offers a new method for the rapid sorting of production line batteries, supporting the enhancement of battery pack consistency and performance.

Key words: production line big data, lithium-ion battery, internal resistance prediction, clustering analysis, battery sorting

中图分类号: