锂电池寿命预测对能源管理和维护具有重要意义,为解决预测过程中复杂的多维时间序列数据、长时间依赖关系以及特征的动态变化等问题,提出一种基于动态卷积神经网络层和Transformer（DCF）、协方差矩阵自适应调整的进化策略（CMA-ES）和多头自注意力机制的锂电池寿命预测模型。DCF通过动态提取时间序列中的关键特征,降低数据维度和冗余性,捕捉长时间依赖;CMA-ES优化模型超参数,增强模型对局部特征与全局依赖的建模能力;多头自注意力机制则进一步聚焦重要特征,处理复杂的非线性动态关系。使用NASA提供的公开锂电池数据集进行实验验证,结果表明该方法的平均绝对误差最小达到0.28%,优于大部分使用同一数据集的现有方法。实验结果进一步证明,模型在预测准确度和泛化能力上均有提升,尤其在长期寿命预测中展现出更高的精度和鲁棒性,可为锂电池的寿命预测提供更为可靠的技术支持。
Abstract
The prediction of lithium battery lifespan is of great significance for energy management and maintenance. To address challenges such as complex multidimensional time series data, long-term dependencies, and dynamic changes in characteristics during the prediction process, this paper proposes a lithium-ion battery life prediction model based on dynamic convolutional neural networks and Transformer（DCF）, covariance matrix adaptive adjustment evolution strategy（CMA-ES）, and multi-head self-attention mechanism. The DCF dynamically extracts key features from time series data, reduces data dimensionality and redundancy, and captures long-term dependencies. CMA-ES optimizes model hyperparameters to enhance the model's ability to model local features and global dependencies. The multi-head self-attention mechanism further focuses on important features and handles complex nonlinear dynamic relationships. Experimental validation is conducted using the publicly available lithium battery dataset provided by NASA. Results show that the proposed method achieves a minimum average absolute error of 0.28%, outperforming most existing methods using the same dataset. The experiments further demonstrate improvements in prediction accuracy and generalization ability, especially in long-term lifespan prediction, where the model exhibits higher precision and robustness, providing more reliable technical support for lithium battery lifespan prediction.
关键词
锂电池 /
卷积神经网络 /
协方差矩阵 /
多头自注意力机制 /
模型超参数 /
非线性
Key words
lithium batteries /
convolutional neural networks /
covariance matrix /
multi head self attention mechanism /
model hyperparameters /
nonlinear
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] YONG J Y, RAMACHANDARAMURTHY V K, TAN K M, et al.A review on the state-of-the-art technologies of electric vehicle, its impacts and prospects[J]. Renewable and sustainable energy reviews, 2015, 49: 365-385.
[2] ARDESHIRI R R, BALAGOPAL B, ALSABBAGH A, et al.Machine learning approaches in battery management systems: state of the art: remaining useful life and fault detection[C]//2020 2nd IEEE International Conference on Industrial Electronics for Sustainable Energy Systems (IESES). Cagliari, Italy, 2020: 61-66.
[3] MAHESHWARI A, PATERAKIS N G, SANTARELLI M, et al.Optimizing the operation of energy storage using a non-linear lithium-ion battery degradation model[J]. Applied energy, 2020, 261: 114360.
[4] HOSSAIN LIPU M S, HUSSAIN A, SAAD M H M, et al. Improved recurrent NARX neural network model for state of charge estimation of lithium-ion battery using pso algorithm[C]//2018 IEEE Symposium on Computer Applications & Industrial Electronics (ISCAIE). Penang, 2018: 354-359.
[5] 刘月峰, 赵光权, 彭喜元. 多核相关向量机优化模型的锂电池剩余寿命预测方法[J]. 电子学报, 2019, 47(6): 1285-1292.
LIU Y F, ZHAO G Q, PENG X Y.A lithium-ion battery remaining using life prediction method based on multi-kernel relevance vector machine optimized model[J]. Acta electronica sinica, 2019, 47(6): 1285-1292.
[6] HOW D N T, HANNAN M A, HOSSAIN LIPU M S, et al. State of charge estimation for lithium-ion batteries using model-based and data-driven methods: a review[J]. IEEE access, 2019, 7: 136116-136136.
[7] 董磊, 赖纪东, 苏建徽, 等. 基于IMAFFRLS-EKF的锂电池在线参数辨识和SOC估计方法[J]. 太阳能学报, 2024, 45(6): 66-74.
DONG L, LAI J D, SU J H, et al.Online parameter identification and SOC estimation of lithium battery based on IMAFFRLS-EKF[J]. Acta energiae solaris sinica, 2024, 45(6): 66-74.
[8] 李谦, 姜帆, 韩乔妮, 等. 基于一阶ECM-IGPR的锂离子电池SOC及SOH联合估计框架[J]. 太阳能学报, 2024, 45(5): 240-250.
LI Q, JIANG F, HAN Q N, et al.Joint first order SOC and SOH estimation frame work for Li-ion battery based on ECM-IGPR[J]. Acta energiae solaris sinica, 2024, 45(5): 240-250.
[9] 魏腾飞, 潘庭龙. 基于改进PSO优化LSTM网络的短期电力负荷预测[J]. 系统仿真学报, 2021, 33(8): 1866-1874.
WEI T F, PAN T L.Short-term power load forecasting based on LSTM neural network optimized by improved PSO[J]. Journal of system simulation, 2021, 33(8): 1866-1874.
[10] 莫易敏, 余自豪, 叶鹏, 等. 基于迁移学习与GRU神经网络结合的锂电池SOH估计[J]. 太阳能学报, 2024, 45(3): 233-239.
MO Y M, YU Z H, YE P, et al.Lithium battery SOH estimation method based on combination of transfer learning and GRU neural network[J]. Acta energiae solaris sinica, 2024, 45(3): 233-239.
[11] CHEN D Q, ZHOU X Z.AttMoE: attention with mixture of experts for remaining useful life prediction of lithium-ion batteries[J]. Journal of energy storage, 2024, 84: 110780.
[12] ZHANG Y, TU L, XUE Z W, et al.Weight optimized unscented Kalman filter for degradation trend prediction of lithium-ion battery with error compensation strategy[J]. Energy, 2022, 251: 123890.
[13] QIAN C, XU B H, CHANG L, et al.Convolutional neural network based capacity estimation using random segments of the charging curves for lithium-ion batteries[J]. Energy, 2021, 227: 120333.
[14] WANG X H, DAI K, HU M, et al.Lithium-ion battery health state and remaining useful life prediction based on hybrid model MFE-GRU-TCA[J]. Journal of energy storage, 2024, 95: 112442.
PDF(1746 KB)
