基于信息熵与PSO-LSTM的锂电池组健康状态估计方法  

State-of-health Estimate for Lithium-ion Battery Using Information Entropy and PSO-LSTM

作　　者：张朝龙 赵筛筛[1] 何怡刚 ZHANG Chaolong;ZHAO Shaishai;HE Yigang(School of Electronic Engineering and Intelligent Manufacturing,Anqing Normal University,Anqing 246011;School of Electrical and Automation,Wuhan University,Wuhan 430072)

机构地区：[1]安庆师范大学电子工程与智能制造学院,安庆246011 [2]武汉大学电气与自动化学院,武汉430072

出　　处：《机械工程学报》2022年第10期180-190,共11页Journal of Mechanical Engineering

基　　金：国家自然科学基金(51637004,51607004);国家重点研发计划(2016YFF0102200);安徽高校协同创新(GXXT-2019-002);安徽高校自然科学研究重点(KJ2020A0509);安庆师范大学研究生学术创新(2021yjs XSCX009)资助项目。

摘　　要：针对目前锂电池组健康状态估计方法的不足,提出一种基于信息熵与粒子群算法(Particle swarm optimization,PSO)优化长短时记忆神经网络(Long short-term memory neural network,LSTM)的锂电池组健康状态估计方法。基于锂电池组恒流-恒压充电阶段锂电池组内各单体端电压的信息熵和平均温度信息,应用PSO-LSTM方法提取锂电池组电压熵、平均温度和锂电池组健康状态之间的映射关系,从而建立锂电池组健康状态估计模型。应用试验室测量的锂电池组老化数据对提出的方法进行测试。测试结果表明,该方法能够准确估计锂电池组的健康状态,平均估计误差在1%以内。同时,为验证提出的方法可推广至锂电池单体,利用美国航天航空局测得的锂电池加速老化数据再次测试,平均估计误差在0.7%以内。并针对锂电池组与锂电池单体设计对比试验,进一步验证提出的方法具有良好的估计性能。In order to address the shortcoming of the existing lithium-ion battery pack state of health(SOH)estimation methods,a SOH estimation approach for lithium-ion battery pack using information entropy and particle swarm optimization(PSO)to optimize the long short-term memory(LSTM)neural network is proposed.The data of the information entropy and the average temperature of each cell terminal voltage in the lithium-ion battery pack during the constant current-constant voltage charging stage are utilized to extract the mapping relationship between the voltage entropy,average temperature,and SOH of the lithium-ion battery pack using PSO-LSTM,and then establish the lithium-ion battery pack SOH estimation model.The measured aging data of lithium-ion battery pack in the laboratory are employed to verify the validity of the presented method.The results show that the developed approach can accurately estimate the SOH of the lithium-ion battery pack with the average estimation error within 1%.Meanwhile,in order to verify the proposed method can be extended to lithium-ion batteries,the accelerated aging data of lithium-ion batteries measured by National Aeronautics and Space Administration(NASA)to test again with the average estimation error within 0.7%.The compared experiment is designed for the battery pack and cells,which further demonstrates that the suggested method offers a favorable estimation performance.

关 键 词：锂电池组 健康状态 信息熵 粒子群算法 长短时记忆神经网络 

分 类 号：TM910[电气工程—电力电子与电力传动]