摘要
电池健康状态(state of health,SOH)的准确估算是保证电动汽车高效安全运行的关键。从电化学阻抗谱(electrochemical impedance spectroscopy,EIS)中提取健康特征可实现电池SOH的准确估算,但在线采集EIS数据对车载设备要求较高不易实现,而采集单频率阻抗作为特征进行SOH估算又面临精度较低的问题。针对该问题,本文提出一种基于组合频率特征的SOH估算方法,首先,通过对实验数据进行分析,将前120次循环的10 Hz虚部和后320次循环7.94 Hz的虚部进行组合,形成电池组合频率阻抗特征。然后,基于组合频率阻抗特征,利用B1和B2电池测试数据建立电池SOH估算的长短期记忆(long short term memory,LSTM)神经网络模型,并通过B3和B4号电池数据对模型进行验证。结果显示,采用组合频率阻抗特征建立的SOH估算模型的均方根误差最小为0.3%,相比采用单频率阻抗特征所建立的模型,其误差减小23.9%以上。由此可见,本文所提出的基于组合频率特征的SOH估算方法,不仅阻抗测量过程简单,且估算精度较高,可应用于电池SOH的在线估算。
Accurate estimating the state of health(SOH)of electric vehicle batteries is crucial for their safe and efficient operation.One approach to achieve high-precision SOH estimation is by extracting health characteristics from electrochemical impedance spectroscopy(EIS).However,collecting online EIS data requires high-tech on-board equipment,posing a challenge to the efficacy of this technique.In addition,SOH estimation based on single frequency impedance leads to low accuracy.To address these issues,a new SOH estimation method has been proposed in this study that combines the frequency impedance characteristics The method involves forming a combination of frequency impedance characteristics by merging the imaginary impedance part in the first 120 cycles at 10 Hz with that in the last 320 cycles at 7.94 Hz after analyzing the experimental data.The method then involved training a long short term memory neural network model with test data from B1 and B2 cells to estimate battery SOH,based on the selected combination frequency impedance characteristics.Subsequently,this model was validated with data from B3 and B4 cells.Results estimate that SOH estimation model based on the combination of frequency impedance features yields a root mean square error of 0.3%at least.This figure is at least 23.9%lower than that achieved with the single frequency impedance model.Therefore,the SOH estimation method not only facilitates performing impedance measurements,but it also promises high estimation accuracy.Additionally,it can be applied to online SOH estimation.
作者
李林泽
张向文
LI Linze;ZHANG Xiangwen(School of Electronic Engineering and Automation,Guilin University of Electronic Technology;Key Laboratory of Intelligence Integrated Automation in Guanxi Universities,Guilin University of Electronic Technology,Guilin 541004,Guangxi,China)
出处
《储能科学与技术》
CAS
CSCD
北大核心
2023年第5期1705-1712,共8页
Energy Storage Science and Technology
基金
国家自然科学基金(62263006)
广西自然科学基金项目(2018GXNSFAA281282)
桂林电子科技大学研究生教育创新计划资助项目(2021YCXS120)。
关键词
锂离子电池
健康状态
电化学阻抗谱
相关性分析
长短期记忆神经网络
lithium-ion battery
state of health
electrochemical impedance spectroscopy
correlation analysis
long short-term memory neural network
