1.College of Aerospace Engineering,Shenyang Aerospace University,Shenyang 110136,China
2.Liaoning General Aviation Academy,Shenyang 110136,China
3.Department of Mechanical and Aerospace Engineering,University of California San Diego,La Jolla 92093,USA
Show less
文章历史+
Received
Accepted
Published
2024-11-13
2025-04-12
2025-08-25
Issue Date
2025-10-30
PDF (1801K)
摘要
锂离子电池的荷电状态(state of charge,SOC)是新能源电动汽车电池管理系统中的关键参数。针对复杂运行环境下锂离子电池SOC预测精度不足等问题,提出了一种基于Transformer神经网络的电动汽车锂离子电池SOC智能预测方法。以日产Leaf电池为研究对象,搭建了新能源电动汽车锂离子电池充放离子电测试平台,模拟用户的真实能量需求及实时能量需求的动态变化,动态调整电池的充放电策略,采集多维度电池数据并进行预处理。构建基于Transformer模型的SOC预测框架,通过神经网络提取复杂时间序列特征,实现了对锂离子电池SOC的高精度预测。实验结果表明,提出的方法在预测精度上优于其他网络,其平均绝对误差低于1.51%,均方根误差(root mean square error,RMSE)低于0.48%,验证了该方法的有效性和准确性。
Abstract
The state of charge (SOC) of lithium-ion batteries is a critical parameter in the battery management system of new energy electric vehicles. To address the issue of insufficient SOC prediction accuracy for lithium-ion batteries under complex operating conditions,an intelligent SOC prediction method for electric vehicle lithium-ion batteries based on the Transformer neural network was proposed. Taking the Nissan Leaf battery as the research object, a charging and discharging test platform for new energy electric vehicle lithium-ion batteries was built to simulate the real energy demands of users and the dynamic changes in real-time energy needs. This platform dynamically adjusted the battery’s charging and discharging strategies, collected multi-dimensional battery data, and preprocessed the data. Then, a SOC prediction framework based on the Transformer model was constructed, which extracted complex time series features through neural networks, achieved high-precision predictions of lithium-ion battery SOC. The experimental results indicate that the proposed method outperforms other networks in prediction accuracy, with a mean absolute error of less than 1.51% and a RMSE of less than 0.48%, validating the effectiveness and accuracy of this method.
锂离子电池的SOC预测方法主要分为基于模型的预测方法[4-5]及数据驱动的预测方法[6-8]。在基于模型的预测方法中,Sun等[9]提出一种基于变窗口自适应卡尔曼滤波的SOC估算方法,通过检测误差信息变化、更新误差矩阵以减小SOC估计误差。Plett[10]提出了一种加权最小二乘法,用于改进因电流和电压等测量噪声导致的准确率低下问题。刘芳等[11]提出了一种改进的遗传算法,用于电池内阻、极化内阻和电容等参数的在线辨识。这种改进方法有效地避免了遗传算法在锂电池模型参数辨识过程中易陷入局部最优解的局限性。Yang等[12]提出了一种基于模型的扩展卡尔曼滤波算法进行SOC估计,该算法结合了在Simulink环境中开发的车辆动力学模型,实现对SOC的精准预测。Pang等[13]提出了一种自适应卡尔曼滤波算法,将卡尔曼滤波与最小二乘法相结合,与传统的固定参数卡尔曼滤波相比,该方法能够显著减小SOC预测结果的偏差。赵剑坤等[14]提出了一种考虑电池衰退影响的动静态SOC估算方法,该方法基于安时积分法进行SOC估算,并通过引入电池健康状态(state of health,SOH)修正电池的初始有效容量。对于静态SOC估算,主要考虑电池衰退因素;而在动态SOC估算中,不仅考虑了电池的衰退,还综合考虑了运行电流和温度的影响。
ZhangW, LiX, LiX.Deep learning-based prognostic approach for lithium-ion batteries with adaptive time-series prediction and on-line validation[J].Measurement,2020,164:108052.
[2]
ZhuY F, LiX, ZhangY W,et al.Cross-domain prognostic method of lithium-ion battery in new energy electric aircraft with domain adaptation[J].IEEE Sensors Journal,2023,23(13):14487-14498.
BuchicchioE, De AngelisA, SantoniF,et al.Battery SOC estimation from EIS data based on machine learning and equivalent circuit model[J].Ener-gy,2023,283:128461.
[5]
DingP, LiuX J, LiH Q,et al.Useful life prediction based on wavelet packet decomposition and two-dimensional convolutional neural network for lithium-ion batteries[J].Renewable and Sustai-nable Energy Reviews,2021,148:111287.
[6]
KorkmazM.SoC estimation of lithium-ion batteries based on machine learning techniques:a filtered approach[J].Journal of Energy Storage,2023,72:108268.
[7]
ZhouH, LuoJ, YuZ.Co-estimation of SOC and SOH for Li-ion battery based on MIEKPF-EKPF fusion algorithm[J].Energy Reports,2023,10(11):4420-4428.
[8]
SangeethaE P, SubashiniN, SanthoshT K,et al.Validation of EKF based SoC estimation using vehicle dynamic modelling for range prediction[J].Electric Power Systems Research,2024,226:109905.
[9]
SunD M, YuX L, WangC M,et al.State of charge estimation for lithium-ion battery based on an Intelligent Adaptive Extended Kalman Filter with improved noise estimator[J].Energy,2021,214:119025.
[10]
PlettG L.Recursive approximate weighted total least squares estimation of battery cell total capa-city[J].Journal of Power Sources,2011,196(4):2319-2331.
YangB, LiG, TangW, et al. Research on optimized SOC estimation algorithm based on extended Kalman filter [J]. Frontiers in Energy Research, 2022, 10: 1027439.
[13]
PangY Z, ZhangX Y, TianJ T. The SOC estimation for power battery using KF method which parameters are updated by least square method [J]. Applied Mechanics and Materials, 2014, 694: 67-72.
ChenP Y, MaoZ Y, WangC Y,et al.A novel RBFNN-UKF-based SOC estimator for automatic underwater vehicles considering a temperature compensation strategy[J].Journal of Energy Sto-rage,2023,72:108373.
[16]
Takyi-AninakwaP, WangS L, LiuG C,et al.Enhanced extended-input LSTM with an adaptive singular value decomposition UKF for LIB SOC estimation using full-cycle current rate and temperature data[J].Applied Energy,2024,363:123056.
[17]
ShaoY X, ZhengY J, ZhangJ N,et al.A cloud capacity estimation method for electric vehicle lithium-ion battery independent of cloud SOC[J].Journal of Energy Storage,2024,85:110998.
[18]
ChenG X, PengW W, YangF F.An LSTM-SA model for SOC estimation of lithium-ion batteries under various temperatures and aging levels[J].Journal of Energy Storage,2024,84:110906.
HuJ X, HuW H, CaoD,et al.Probabilistic net load forecasting based on transformer network and Gaussian process-enabled residual modeling lear-ning method[J].Renewable Energy,2024,225:120253.
[22]
ZhangH L, TangW, NaW,et al.Implementation of generative adversarial network-CLS combined with bidirectional long short-term memory for lithium-ion battery state prediction[J].Journal of Energy Storage,2020,31:101489.
[23]
GuoX Y, LinX, YangX,et al.UCTNet:Uncertainty-guided CNN-Transformer hybrid networks for medical image segmentation[J].Pattern Recognition,2024,152:110491.
[24]
LiX, ZhangW, LiX,et al.Partial domain adaptation in remaining useful life prediction with incomplete target data[J].IEEE/ASME Transactions on Mechatronics,2024,29(3):1903-1913.
[25]
YangS, LingL, LiX, et al. Industrial battery state-of-health estimation with incomplete limited data towards second-life applications [J]. Journal of Dynamics, Monitoring and Diagnostics, 2024,3(4):246-257.