第24卷第2期2023年2月电气技术ElectricalEngineeringVol.24No.2Feb.2023相关向量机预测锂离子电池剩余有效寿命余佩雯1,2郁亚娟1,2常泽宇1张之琦1陈来1,2(1.北京理工大学材料学院,北京100081;2.北京理工大学重庆创新中心,重庆401120)摘要随着新能源汽车的迅速发展,锂离子电池已得到广泛应用。准确预测锂离子电池剩余有效寿命(RUL)对于合理规划电池使用至关重要。目前,机器算法和模型预测已广泛应用于电池剩余有效寿命的预测中。本文基于数据驱动的方法进行锂离子电池剩余有效寿命预测,通过使用相关向量机(RVM)将长期预测分为多段短期预测,并结合自相关函数、灰色关联度模型、卡尔曼滤波器(KF)进行模型优化与改进,改进后的RVM模型在三组目标电池RUL预测中的相对误差分别为5.46%、7.14%和6.29%,与其他几种预测模型的对比结果表明该模型优于其他模型。关键词:剩余有效寿命(RUL);锂离子电池;相关向量机(RVM);灰色关联度模型Remainusefullifepredictionoflithium-ionbatterybasedonrelevancevectormachineYUPeiwen1,2YUYajuan1,2CHANGZeyu1ZHANGZhiqi1CHENLai1,2(1.SchoolofMaterialsScience&Engineering,BeijingInstituteofTechnology,Beijing100081;2.BeijingInstituteofTechnologyChongqingInnovationCenter,Chongqing401120)AbstractWiththerapiddevelopmentofnewenergyvehicles,lithium-ionbatterieshavebeenwidelyused.Accuratelypredictingitsremainingusefullife(RUL)iscrucialforrationalplanningofbatteryusage.Atpresent,machinealgorithmsandmodelpredictionhavebeenwidelyusedtopredictthebatteryremainingusefullife.Thisstudyadoptsadata-drivenmethodtopredicttheRULoflithium-ionbattery.Byusingtherelevancevectormachine(RVM),thelong-termforecastisdividedintomultipleshort-termforecasts,whichiscombinedwithauto-correlationfunction,greycorrelationmodel,Kalmanfilter(KF)tooptimizeandimprovethemodel.TherelativeerrorsofthepredictionbasedonthemodifiedRVMmodelinthethreegroupoftargetcellsare5.46%,7.14%,and6.29%,respectively.Theresultsshowthatthepredictionresultsofthismodelarebetterthanothermodels.Keywords:remainusefullife(RUL);lithium-ionbattery;relevancevectormachine(RVM);greycorrelationmodel0引言近年来,锂离子电池在新能源汽车领域得到了广泛应用[1]。锂离子电池凭借...