基于弛豫时间分布法的退役动力电池健康状态评估  

SOH estimation based on distribution of relaxation times for the retired power lithium-ion battery

作　　者：张子恒 耿萌萌 范茂松 金玉红 刘晶冰[1] 杨凯 汪浩[1] ZHANG Ziheng;GENG Mengmeng;FAN Maosong;JIN Yuhong;LIU Jingbing;YANG Kai;WANG Hao(College of Materials Science and Engineering Beijing University Of Technology,Beijing 100124,China;China Electric Power Research Institute,Beijing 100192,China)

机构地区：[1]北京工业大学材料科学与工程学院,北京100124 [2]中国电力科学研究院有限公司,北京100192

出　　处：《储能科学与技术》2025年第2期770-778,共9页Energy Storage Science and Technology

基　　金：国家电网有限公司总部管理科技项目(5108-202218280A-2-314-XG)。

摘　　要：退役电池在进行梯次利用之前需要对其参数进行测试与评估,以保证后续依据电池的性能为其选择适合的应用场景。健康状态(state of health,SOH)的准确评估是退役动力电池是否有梯次利用价值的前提。针对目前退役动力电池SOH评估准确性低的问题。本工作首先利用弛豫时间分布法对电化学阻抗谱进行分析,用以得到其中能准确反应电池健康状态的特征频率,将特征频率对应的阻抗数据作为特征输入参量,输入麻雀算法优化的极限学习机模型以实现退役动力电池SOH的评估。为了验证评估方法的有效性,针对7只方形磷酸铁锂退役电池进行循环老化实验,并在每个循环周期后进行电化学阻抗测试。使用退役动力电池的实际的电化学阻抗谱用于分析与建模实现SOH评估,并将结果与实际的SOH数据进行对比,并与传统的SOH评估方法进行了对比。评估结果表明,相较于其他方法采用弛豫时间分布法进行分析的均方误差(mean square error,MSE)与平均绝对百分比误差(mean absolute percentage error,MAPE)更低。相较于使用未优化的极限学习机模型,MSE和MAPE分别降低了47.1%和60.5%,表明本文的SOH评估方法具有更小的误差和更高的准确性,在实际的梯次利用中很有应用价值。Retired batteries should undergo rigorous testing and evaluation to ascertain their performance before they are deployed in echelons utilization to ensure that appropriate application scenarios are selected based on the performance of the batteries.An accurate assessment of the State of Health(SOH)is fundamental to determine whether a power battery possesses the value of echelons utilization.In view of the low accuracy of SOH evaluation of retired power batteries,the relaxation time distribution method was used to analyze the electrochemical impedance spectroscopy in this study.This was done to obtain the characteristic frequency that can accurately reflect the health state of the battery.The impedance data corresponding to the characteristic frequency was used as the characteristic input parameters,and the extreme learning machine model optimized using the sparrow algorithm served as input to realize the SOH evaluation of decommissioned power batteries.To verify the effectiveness of the evaluation method,seven retired prismatic lithium-iron-phosphate batteries were subjected to cyclic aging experiments,and electrochemical impedance spectroscopy tests were performed on the batteries after each cycle.Actual electrochemical impedance spectroscopy of the decommissioned power batteries was used for analysis and modeling to evaluate the SOH,and the results were compared with actual SOH data obtained using traditional SOH evaluation methods.Our findings demonstrate that the Mean Square Error(MSE)and Mean Absolute Percentage Error(MAPE)associated with the relaxation time distribution method are lower than those of other methods.Compared with the unoptimized extreme learning machine model,the values of the MSE and MAPE reduced by 47.1%and 60.5%,respectively,indicating that the SOH evaluation method employed in this study is relatively highly accurate and less prone to error,indicating its applicability in practical echelons utilization.

关 键 词：退役动力锂离子电池 健康状态 交流阻抗谱 弛豫时间分布 极限学习机 

分 类 号：TM912[电气工程—电力电子与电力传动]