基于ABC-LSTM模型的锂离子电池剩余使用寿命预测  

Remaining useful life prediction of lithium-ion battery based on an ABC-LSTM model

作　　者：刘勇 于怀汶 刘大鹏 穆勇 王瀛洲[2] 张秀宇[2] LIU Yong;YU Huaiwen;LIU Dapeng;MU Yong;WANG Yingzhou;ZHANG Xiuyu(State Grid Jibei Electric Power Co.,Ltd.Tangshan Power Supply Company,Tangshan 130033,Hebei,China;School of Automation Engineering,Northeast Electric Power University,Jilin 132012,Jilin,China)

机构地区：[1]国网冀北电力有限公司唐山供电公司,河北唐山130033 [2]东北电力大学自动化工程学院,吉林吉林132012

出　　处：《储能科学与技术》2025年第1期331-345,共15页Energy Storage Science and Technology

基　　金：国家自然科学基金青年基金(62203102);国网冀北电力有限公司唐山供电公司科技项目(B30103230015);火电机组高效清洁运行智能控制实验平台建设(2020C022-3)。

摘　　要：为了保证储能系统的安全稳定运行,准确预测锂离子电池的剩余使用寿命(remaining useful life,RUL)至关重要。本工作提出了一种基于人工蜂群算法(artificial bee colony,ABC)和结合dropout技术的长短期记忆网络(long short-term memory,LSTM)相结合的综合预测模型,可有效提高锂离子电池RUL预测的准确性。首先,利用dropout正则化方法有效减轻过拟合现象的优势,提高预测模型的泛化能力。其次,引入针对容量回升及数据噪声问题的激活层网络结构,显著提升模型对复杂非线性数据的处理能力。然后,结合ABC算法优化LSTM综合预测模型的超参数,避免模型陷入局部最优解,提高RUL预测精度。最后,通过NASA研究中心及CALCE的公开数据集验证所提模型的预测准确性和鲁棒性。本工作对基于40%和60%训练数据的不同算法预测性能进行实验分析验证,并与麻雀优化算法、座头鲸优化算法等群体优化算法进行比较。实验结果表明,所提出的ABC-LSTM综合预测模型可以更加准确地捕获锂离子电池容量退化的全局趋势及局部特征,其中60%比例的RUL预测结果的均方根误差平均保持在1.02%以内,平均绝对误差平均保持在0.86%以内,拟合系数高达97%以上。To ensure the safe and stable operation of energy storage systems,the accurate prediction of the remaining useful life(RUL)of lithium-ion batteries is crucial.This study presents an integrated forecasting model that combines the artificial bee colony(ABC)algorithm with a long short-term memory(LSTM)network enhanced by dropout techniques.This combination effectively improves the accuracy of RUL predictions for lithium-ion batteries.First,the dropout regularization method is utilized to effectively mitigate overfitting,thereby enhancing the generalization capability of the predictive model.Subsequently,an activation layer network structure is introduced to address capacity recovery and data noise issues,significantly enhancing the ability of the model to handle complex nonlinear data.Thereafter,the hyperparameters of the LSTM-based comprehensive forecasting model are optimized using the ABC algorithm to avoid local optima and improve the precision of RUL predictions.Finally,the predictive accuracy and robustness of the proposed model are verified using a public dataset from the NASA Research Center and the CALCE.The paper conducts an experimental analysis and verification of The predictive performance of various algorithms were experimentally analyzed and verified using training data at 40%and 60%levels.The performance of swarm optimization algorithms,such as the Sparrow Search Algorithm and the Humpback Whale Optimization Algorithm,were also compared.The experimental results demonstrate that the proposed ABC-LSTM integrated forecasting model can capture the global trends and local characteristics of the capacity degradation of the lithium-ion battery more accurately than the compared models.The root-mean-squared error of the RUL prediction results obtained with a 60%proportion of training data remained consistently within 1.02%;the mean absolute error remained consistently within 0.86%,and the fitting coefficient exceeded 97%.

关 键 词：锂离子电池 剩余使用寿命预测 长短期记忆网络 人工蜂群算法 dropout技术 

分 类 号：TK912[动力工程及工程热物理]