《电子技术应用》
您所在的位置:首页 > 电源技术 > 设计应用 > 基于FNN的锂电池SoC预测
基于FNN的锂电池SoC预测
集成电路应用
徐志伟,于莲芝
上海理工大学光电信息与计算机工程学院
摘要: 针对锂电池的安全性与稳定性的核心问题,该文提出使用前馈神经网络(FNN)对锂电池参数进行预测。首先利用等效电路图和电压电流变化来分析锂电池参数特点和内部模型函数,随后采用FNN对锂电池参数进行预测。为了进一步提高预测精度并减小误差,在不同温度环境下进行预测,并得到相对的均方根误差(RMSE)。试验结果表明,FNN相较于传统的安时积分法、开路电压法等,具有精度高,误差小等显著优势。
中图分类号:TM912文献标识码:ADOI:10.19339/j.issn.1674-2583.2026.07.003
中文引用格式:徐志伟,于莲芝. 基于FNN的锂电池SoC预测[J].集成电路应用,2026,43(7):12-19.
英文引用格式:Xu Zhiwei, Yu Lianzhi. Lithium battery SoC prediction based on FNN[J].Application of IC,2026,43(7):12-19.
Lithium battery SoC prediction based on FNN
Xu Zhiwei, Yu Lianzhi
School of Optoelectronic Information and Computer Engineering, University of Shanghai for Science and Technology
Abstract: Addressing the core concerns regarding the safety and stability of lithium batteries, this study proposes a Feedforward Neural Network (FNN) for predicting lithium battery parameters. Initially, the characteristics of lithium battery parameters and internal model functions are analyzed using equivalent circuit diagrams and voltage-current variations, leading to the selection of the FNN for parameter prediction. To further enhance prediction accuracy and to reduce errors, predictions were conducted under various temperature conditions, yielding relative Root Mean Squared Error (RMSE) values. Experimental results indicate that, compared with traditional methods such as the ampere-hour integration method and open-circuit voltage method, FNN achieves high accuracy with minimum error.
Key words : lithium battery; equivalent circuit voltage model; terminal voltage curve; lithium battery parameters; FNN; optimization algorithm

引言

本文针对锂电池安全问题对锂电池荷电状态(State of Charge,SoC)进行预测,使用前馈神经网络(Feedforward Neural Network,FNN)作为预测算法,可以有效得到预测结果与误差,实现不同温度下的锂电池SoC预测,为电动汽车、储能电站等应用场景所需要的锂电池安全和性能预测提供可靠的方法。


本文详细内容请下载:

http://www.chinaaet.com/resource/share/2000007310


投稿邮箱:ICAPP@belling.com.cn


作者信息:

徐志伟,于莲芝

(上海理工大学光电信息与计算机工程学院,上海200093)

2.jpg

此内容为AET网站原创,未经授权禁止转载。