储能科学与技术 ›› 2023, Vol. 12 ›› Issue (2): 602-608.doi: 10.19799/j.cnki.2095-4239.2022.0403

• 储能测试与评价 • 上一篇    下一篇

基于GAPSO-FNN神经网络的锂离子电池传感器故障诊断

王志福1,2(), 罗崴2, 闫愿1, 徐崧1, 郝文美1, 周聪林3   

  1. 1.北京理工大学机械与车辆学院，北京 100081
    2.广西科技大学自动化学院，广西 柳州 545000
    3.上海电机学院商学院，上海 201306
  • 收稿日期:2022-07-18 修回日期:2022-12-08 出版日期:2023-02-05 发布日期:2023-02-24
  • 通讯作者: 王志福 E-mail:wangzhifu@bit.edu.cn
  • 作者简介:王志福（1977—），男，博士，教授，从事智能电动车动力学等研究，E-mail：wangzhifu@bit.edu.cn。
  • 基金资助:
    国家自然科学基金项目(51775042)

Fault diagnosis of lithium-ion battery sensors using GAPSO-FNN

Zhifu WANG1,2(), Wei LUO2, Yuan YAN1, Song XU1, Wenmei HAO1, Conglin ZHOU3   

  1. 1.School of Machinery and Vehicles, Beijing Institute of Technology, Beijing 100081, China
    2.School of Automation, Guangxi University of Science and Technology, Liuzhou 545000, Guangxi, China
    3.Business School of Shanghai Electric Machinery University, Shanghai 201306, China
  • Received:2022-07-18 Revised:2022-12-08 Online:2023-02-05 Published:2023-02-24
  • Contact: Zhifu WANG E-mail:wangzhifu@bit.edu.cn

RichHTML

31

PDF

176

摘要：

新能源汽车的动力电池内部存在多种传感器用来进行电池系统的安全监测，而传感器故障会导致荷电状态等指标出现误差，严重时会触发电池热失控的风险。为了有效准确地进行电池传感器故障诊断，提出基于遗传算法优化粒子群算法(genetic algorithm optimized particle swarm optimization，GAPSO)和模糊神经网络(fuzzy neural network， FNN)的锂离子电池传感器故障诊断方法对锂离子电池的传感器进行故障诊断，该方法使识别故障准确率迅速提升。本工作首先通过硬件平台和Matlab/Simulink环境相结合的方式获取电池传感器故障的数据，然后对故障数据进行预处理及特征提取，最后采用GAPSO-FNN算法对电池传感器进行故障诊断，并与传统神经网络和模糊神经网络方法的结果进行对比。仿真结果表明，基于GAPSO-FNN的锂离子电池传感器故障诊断方法相比于传统的神经网络方法测量准确率提升了25%，相比于模糊神经网络准确率提升了10%，故障诊断准确率能够达到95%，在减少故障诊断所需信息量的同时，有效地提升了故障诊断的准确率。

关键词: 锂离子电池, 传感器故障诊断, GAPSO-FNN, 健康监测, 热失控风险

Abstract:

Various sensors inside the power battery of novel energy vehicles are used to monitor the safety of the battery system, and sensor failure can cause errors in the charge state and other indicators, which can severely trigger the risk of thermal runaway. For effective and accurate battery sensor fault diagnosis, we propose a fault diagnosis method based on genetic algorithm-based particle swarm optimization and fuzzy neural network (GAPSO-FNN). The proposed method is applied to diagnose the sensor faults of lithium-ion batteries. We obtained data on battery sensor faults by combining the hardware platform and Matlab/Simulink environment, then preprocessed and extracted features from the fault data, and finally used the GAPSO-FNN-based method to diagnose battery sensor faults and compared the results with conventional neural network (NN)- and fuzzy neural network (FNN)-based methods. Simulation results show that the GAPSO-FNN-based method improves the accuracy by 25% and 10% compared with the conventional NN- and FNN-based methods and the fault diagnosis accuracy can reach 95%. Thus, the proposed method effectively improves fault diagnosis accuracy while reducing the amount of information required for fault diagnosis.

Key words: lithium-ion battery, sensor fault diagnosis, GAPSO-FNN, health monitoring, thermal runaway risk

中图分类号: