摘要
锂电池荷电状态(SOC)的精准预测对电池安全运行具有关键作用,针对传统的锂电池SOC难以精准预测的问题,本文提出了一种基于变分模式分解(VMD)、卷积神经网络(CNN)和长短时记忆网络(LSTM)融合的VMD-CNN-LSTM预测模型。首先采用VMD分解原始数据,得到一系列具有不同频率特性的子信号,然后通过CNN提取锂电池数据中不同维度数据间的特征关系,再经过LSTM提取长期特征,并将模态分量的预测结果重构,实现SOC的高精度预测。实验结果表明,该VMD-CNN-LSTM模型在锂电池SOC预测中取得了较高的精度,明显优于传统的神经网络模型预测方法。该研究不仅为锂电池SOC预测提供了一种新的有效方法,也为新能源锂电池的发展提供了有力支持。
The accurate prediction of state of charge(SOC)of lithium battery plays a key role in the safe operation of the battery.Aiming at the difficulty of accurate prediction of traditional SOC of lithium battery,a VMD-CNN-LSTM pre⁃diction model based on the fusion of variational mode decomposition(VMD),convolutional neural network(CNN)and long and short term memory network(LSTM)is proposed in this paper.Firstly,the original data is decomposed by VMD to obtain a series of sub-signals with different frequency characteristics.Then,the feature relationship between the data of different dimensions of lithium battery data is extracted through CNN,and long term features are extracted through LSTM,and the prediction results of modal components are reconstructed to achieve high precision prediction of SOC.The experimental results show that the VMD-CNN-LSTM model achieves high accuracy in SOC prediction of lithium bat⁃tery,which is obviously superior to the traditional neural network model prediction method.This study not only provides a new and effective method for SOC prediction of lithium batteries,but also provides strong support for the development of new energy lithium batteries.
作者
王飞
吕明琪
Wang Fei;LüMingqi(College of Control Science and Engineering,China University of Petroleum(East China),Qingdao 266000,China;Zhejiang University of Technology Artificial Intelligence Innovation Research Insti-tute,Hangzhou 310056,China;School of Computer Science,Zhejiang University of Technology,Hangzhou 310000,China)
出处
《电力电子技术》
2025年第6期67-71,共5页
Power Electronics
基金
杭州市重点科研计划(2024SZD0220)。
关键词
锂电池
荷电状态
变分模式分解
卷积神经网络
长短时记忆网络
lithium battery
state of charge
variational mode decomposition
convolutional neural network
long and short term memory network
