基于神经网络的磷酸铁锂电池SOC预测研究.pdfVIP

第 25 卷 第 5 期 电子测量与仪器学报 Vol. 25 No. 5 2011 年 5 月 JOURNAL OF ELECTRONIC MEASUREMENT AND INSTRUMENT · 433 · DOI: 10.3724/SP.J.1187.2011.00433 基于神经网络的磷酸铁锂电池 SOC 预测研究* 1,2 1 1,2 1,2 尹安东 张万兴 赵 韩 江 昊 (1. 合肥工业大学机械与汽车工程学院, 合肥 230009; 2. 安徽省汽车技术与装备工程研究中心, 合肥 230009) 摘 要: 利用神经网络进行了电动汽车用的磷酸铁锂(LiFePO ) 电池荷电状态(state of charge, SOC)预测研究。在分析磷酸 4 铁锂电池充放电机理的基础上, 采用 levenberg-marquardt (LM)算法建立了磷酸铁锂电池的 BP (back propagation)神经网络模型, 并进行了电池 SOC 值的预测。结果表明, 基于神经网络的电池 SOC 预测方法具有较高的精度, 可用来预测磷酸铁锂电池的 SOC 值。 关键词: 电动汽车；磷酸铁锂电池；荷电状态(SOC) ；神经网络；预测 中图分类号: TM912 文献标识码: A 国家标准学科分类代码: 510.4130 Research on estimation for SOC of LiFePO Li-ion battery based on neural network 4 Yin Andong1,2 Zhang Wanxing1 Zhao Han1,2 Jiang Hao1,2 (1. School of Machinery and Automobile Engineering, Hefei University of Technology, Hefei 230009, China; 2. Anhui Research Center for Automotive Technology Equipment Engineering, Hefei 230009, China) Abstract: The state of charge (SOC) of LiFePO Li-ion battery for electric vehicle was estimated by neural network. 4 Based on the electrochemical mechanism of LiFePO4 Li-ion batteries during charge and discharge, a back propagation (BP) neural network model was built up using Levenberg-Marquardt(LM). The neural network model was used to estimate SOC of LiFePO4 Li-ion battery. The results show that the neural network method is a quite accurate algorithm and can be used in estimation of SOC of LiFePO4 Li-ion batteries. Keywords: electric vehicle; LiFePO4 Li-ion battery; state of charge(SOC); ne