基于一致性检验的锂离子电池组健康状态预测  

Prediction of lithium-ion battery pack SOH based on consistency check

作　　者：许万里 洪小波 王东阳 阮殿波 XU Wanli;HONG Xiaobo;WANG Dongyang;RUAN Dianbo(College of Mechanical Engineering and Mechanics,Ningbo University,Ningbo Zhejiang 315211,China;Institute of Advanced Energy Storage Technology and Equipment,Ningbo University,Ningbo Zhejiang 315211,China)

机构地区：[1]宁波大学机械工程与力学学院,浙江宁波315211 [2]宁波大学先进储能技术与装备研究院,浙江宁波315211

出　　处：《电源技术》2025年第4期772-781,共10页Chinese Journal of Power Sources

基　　金：浙江省科技计划项目(2022C01072)。

摘　　要：准确预测锂离子电池的健康状态(SOH)对确保电池系统的安全可靠运行至关重要,然而,电池组的老化与电池内部的电化学反应及单体间的不一致性相关。因此,提出一种基于一致性检验的锂离子电池组SOH预测方法。首先,从单体的恒流充电电压曲线中提取多个特征,依据皮尔逊相关法和不同特征组合对单体SOH进行估计,确定最优特征集合,并建立单体SOH模型;其次,在2并20串电池组上使用一致性检验方法,设立不同阈值筛选离群单体;最后,提取离群单体集合内各单体特征,结合单体SOH模型,预测电池组全生命周期的SOH。结果表明,仅需20%的单体数据,电池组SOH预测的均方根误差(RMSE)可达到0.64%,相比未使用一致性检验方法,RMSE减小5.9%,并且不同老化状态下的SOH预测的相对误差均在1.5%以内。Accurate prediction of the state of health(SOH)of lithium-ion batteries is essential to ensure the safe and reliable operation of battery systems.However,the aging of the battery pack is related to the electrochemical reactions within the battery and the inconsistencies between the batteries.Therefore,this paper proposed a SOH prediction method for lithium-ion battery pack based on consistency check.Firstly,multiple features were extracted from the constant current charging voltage curve of the single battery,and based on Pearson correlation method and estimation effect of different feature combinations on battery SOH,the optimal feature set was determined,and the SOH model was established.Secondly,the consistency check method was used on 2 parallel and 20 series battery pack,and different thresholds were set to screen outlier batteries.Finally,the characteristics of each battery in the outlier batteries set were extracted,and combined with the battery SOH model,the SOH of the whole life cycle of the battery pack was predicted.The results show that the rootmean-square error(RMSE)of SOH prediction of battery pack can reach 0.64%with only 20%single battery data,and RMSE decreases by 5.9%compared with no consistency check method,and the relative error of SOH prediction under different aging states is less than 1.5%.

关 键 词：锂离子电池组 离群单体 健康状态预测 一致性检验 特征集合 

分 类 号：TM912.9[电气工程—电力电子与电力传动]