基于灰色关联度分析-长短期记忆神经网络的锂离子电池健康状态估计

State of Health Estimation for Lithium-Ion Battery Based on Gray Correlation Analysis and Long Short-Term Memory Neural Network

电池的健康状态是电池健康管理的核心,准确的锂离子电池健康状态估计对保证电池安全、可靠、长寿命运行具有重要意义。为此,该文提出了一种基于增量容量曲线和灰色关联度分析(GRA)以及长短期记忆(LSTM)神经网络的锂离子电池健康状态估计方法。该方法通过分析电池在老化过程中的充电增量容量曲线变化模式,提取电池老化特征。为了降低计算复杂度,引入灰色关联度分析法进行特征分析与筛选,并将其作为长短时间记忆神经网络的输入,进行网络预训练进而估计电池的健康状态。最后,利用三种不同工况的电池加速老化测试数据集对所提出的健康状态估计方法进行了验证。实验结果表明,所提出的方法表现出优秀的电池健康状态估计性能,并在不同工况以及不同训练循环周期数条件下表现出良好的鲁棒性。

The state of health is the core of battery health management.An accurate estimation of the state of health is of great significance to ensure the safe,reliable,and long-life operation of lithiumion batteries.To this end,this paper proposes a method for estimating the state of health of lithium-ion batteries based on incremental capacity curves and gray correlation analysis long short-term memory neural networks.This method extracts several different aging characteristics by analyzing the attenuation mode of battery charging incremental capacity curves during the aging process.In order to reduce the computational complexity,the gray correlation analysis method is introduced for features analysis and screening.Then,the extracted aging features are used as the input to train the long short-term memory network and estimate the battery health status.Finally,accelerated battery aging tests based on three different working conditions are conducted to verify the proposed method.The experimental results show that the proposed method exhibits excellent performance in estimating the state of health of the battery,and it shows good robustness under different working conditions and different number of training cycles.