储能科学与技术 ›› 2020, Vol. 9 ›› Issue (4): 1153-1158.doi: 10.19799/j.cnki.2095-4239.2020.0075

• 储能系统与工程 • 上一篇    下一篇

基于粒子群算法的最小二乘支持向量机电池状态估计

王语园1(), 李嘉波2, 张福3   

  1. 1.陕西铁路工程职业技术学院，陕西 渭南 714000
    2.长安大学公路养护装备国家工程实验室，陕西 西安 710064
    3.乌鲁木齐局有限责任公司乌鲁木齐车辆段，新疆 乌鲁木齐 830023
  • 收稿日期:2020-02-19 修回日期:2020-03-13 出版日期:2020-07-05 发布日期:2020-06-30
  • 作者简介:王语园(1981—)，男，硕士，副教授，研究方向为电力电子与电力传动，E-mail：38502042@qq.com。
  • 基金资助:
    渭南市2019年重点研发科技计划项目(2019-ZDYF-JCYJ-127);陕西铁路工程职业技术学院供电科技创新团队(KJTD201901);陕西铁路工程职业技术学院2019年中青年科技创新人才培育项目(KJRC201905)

Battery state estimation of least squares support vector machinebased on particle swarm optimization

WANG"Yuyuan1(), LI"Jiabo2, ZHANG"Fu3   

  1. 1.Shaanxi Railway Institute, Weinan 714000, Shaanxi, China
    2.Highway Maintenance Equipment National Engineering Laboratory, Chang'an University, Xi’an 710064, Shaanxi, China
    3.Urumqi Depot of Urumqi Bureau Co. Ltd. , Urumqi 830023, Xinjiang, China
  • Received:2020-02-19 Revised:2020-03-13 Online:2020-07-05 Published:2020-06-30

RichHTML

20

PDF

250

摘要：

电池荷电状态(SOC)作为电池管理系统(BMS)重要参数之一，准确估计SOC尤为重要。由于SOC在估计过程中常会受到电压、电流、充放电效率等众多因素的影响，因此很难准确估计SOC。为了提高SOC的估计精度，本工作提出了基于最小二乘支持向量机(LSSVM)机器学习的锂离子电池SOC估计模型。将该电池的电流、电压和温度作为模型的输入向量，SOC作为模型的输出向量，为了更好的获得LSSVM模型的参数，提出了利用自适应粒子群算法来进行参数优化，从而获得高精度SOC估计模型。通过恒流充放电实验采集的数据，并和未优化的粒子群优化的LSSVM、支持向量机(SVM)神经网络(NN)相比，所提模型的SOC估计精度误差为1.63%，验证了算法的有效性。

关键词: 锂离子电池, SOC, 最小二乘支持向量机(LSSVM), 粒子群算法

Abstract:

As one of the important parameters of battery management system (BMS), it is very important to estimate SOC accurately. It is difficult to estimate SOC accurately because SOC is often affected by many factors such as voltage, current, charge discharge efficiency and so on. In order to improve the accuracy of SOC estimation, the least square support vector machine (LSSVM) based SOC estimation model for lithium-ion battery is proposed. The current, voltage and temperature of the battery are taken as the input vector and SOC as the output vector of the model. In order to better obtain the parameters of LSSVM model, an adaptive particle swarm optimization algorithm is proposed to optimize the parameters, so as to obtain a high-precision SOC estimation model. Compared with the PSO optimized LSSVM and support vector machine (SVM) neural network (NN), the accuracy error of SOC estimation of the proposed model is 1.63%, which proves the effectiveness of the algorithm.

Key words: lithium-ion battery, SOC, least squares support vector machine (LSSVM), particle swarm optimization

中图分类号: