摘要
探讨了基于外部输入循环神经网络(NARX)在变工况下间接估计锂离子电池荷电状态的方法。先搭建电池等效电路模型,设计NARX神经网络随电池状态切换准确预测极化电压响应的训练工况,然后设计两种不同输入类型NARX神经网络,在DST和UN/ECE(Elementary Urban Cycle)两种测试工况下,进行了极化电压和SOC估计,并将估计数据与前馈神经网络直接估计数据进行比较。数据表明NARX神经网络在变工况下间接估计电池SOC有较高精度。
An indirect estimation method of lithium ion battery SOC based on external input recurrent neural network(NARX)was discussed.First,the equivalent circuit model of the battery was built,and the NARX neural network was designed to accurately predict the training condition of polarization voltage response with the battery state switch.Then,two different input types of NARX neural networks were designed.Under the two test conditions of DST and UN/ECE(Elementary Urban Cycle),the polarization voltage and SOC estimates were performed and compared with the feedforward neural network direct estimates.The data shows that NARX neural network has high accuracy in indirect estimation of battery SOC under variable working conditions.
作者
徐鹏
王潺
万世斌
但远宏
XU Peng;WANG Chan;WAN Shibin;DAN Yuanhong(School of Electrical and Electronics Engineering,Chongqing University of Technology,Chongqing 400054,China;School of Computer Sicence and Engineering,Chongqing University of Technology,Chongqing 400054,China)
出处
《电源技术》
CAS
北大核心
2022年第10期1161-1166,共6页
Chinese Journal of Power Sources
关键词
NARX神经网络
锂离子电池
荷电状态
变工况
NARX neural network
Li-ion battery
state of charge
varying working profile
