基于随机充电数据的锂离子电池容量在线估计  

Online Lithium-ion Battery Capacity Estimation Based on Random Charging Data

作　　者：谷平维 段彬[1] 康永哲 张承慧[1] 杜春水[1] GU Pingwei;DUAN Bin;KANG Yongzhe;ZHANG Chenghui;DU Chunshui(School of Control Science and Engineering,Shandong University,Jinan 250061)

机构地区：[1]山东大学控制科学与工程学院,济南250061

出　　处：《机械工程学报》2023年第22期100-110,共11页Journal of Mechanical Engineering

基　　金：国家自然科学基金资助项目(U1964207,62133007,61821004)。

摘　　要：动力电池是电动汽车的核心部件,其容量的准确在线估计对于保证电动汽车的安全稳定运行至关重要。结合老化特征与数据驱动算法实现电池容量估计是当下的研究热点,然而,电动汽车充放电工况随机,导致难以稳定获取特定的老化特征。为此,提出一种基于随机充电数据的电池容量在线估计方法。首先通过分析不同老化电池的充电数据,提取随机充电片段的累计电量、累计能量、电量均值与方差作为老化特征。然后,基于皮尔逊相关系数和灰色关联度分析法评价所提取的特征与电池容量之间的相关性。最后,提出基于改进核极限学习机的电池容量在线估计方法。与其他数据驱动方法相比,所提出的方法估计精度更高,在任意电压窗口下的容量估计误差平均值为1.38%,特别是在最优窗口下误差仅为0.48%。对比结果表明,所构建的估计模型能够在不同采样频率下保持稳定,同时仅需要少量的新样本即可实现不同充电工况下电池容量的准确估计,所提出的方法泛化性能强,实用价值高。As the core component of electric vehicles,accurate online estimation of lithium-ion battery capacity is crucial to ensure the safe and stable operation of electric vehicles.Combining aging characteristics and data-driven algorithms to achieve battery capacity estimation is a research hotspot.However,the charging and discharging conditions of electric vehicles are random,which makes it difficult to obtain specific aging characteristics stably.To this end,an online battery capacity estimation method based on random charging data is proposed.First,the cumulative electric quantity,cumulative energy,average value and variance of electric quantity of random charging segments are extracted as features by analyzing the charging data of different aging batteries.Then,the correlation between the extracted features and the battery capacity is evaluated based on the Pearson correlation coefficient and the grey correlation degree analysis method.Finally,an online battery capacity estimation method based on improved kernel extreme learning machine is proposed.Compared with other data-driven methods,the proposed method has higher estimation accuracy,with an average capacity estimation error of 1.38%under random voltage window.In particular,the error is only 0.48%under the optimal window.The comparison results show that the constructed estimation model is stable under different sampling frequencies.Meanwhile,only a small number of new samples can be used to accurately estimate the battery capacity under different charging conditions.The proposed method has strong generalization performance and high practical value.

关 键 词：锂离子电池 容量估计 随机充电数据 数据驱动 极限学习机 

分 类 号：TM912[电气工程—电力电子与电力传动]