第47卷第2期电网技术Vol.47No.22023年2月PowerSystemTechnologyFeb.2023文章编号：1000-3673（2023）02-0859-09中图分类号：TM721文献标志码：A学科代码：470·40聚类分析架构下基于遗传算法的电池异常数据检测方法马速良1，武亦文1，李建林1，周琦2，李雅欣1（1．储能技术研究中心(北方工业大学)，北京市石景山区100144；2．国网江苏省电力有限公司电力科学研究院，江苏省南京市211103）AnomalyDetectionforBatteryDataBasedonGeneticAlgorithmUnderClusterAnalysisFrameworkMASuliang1,WUYiwen1,LIJianlin1,ZHOUQi2,LIYaxin1(1.EnergyStorageTechnologyEngineeringResearchCenter(NorthChinaUniversityofTechnology),ShijingshanDistrict,Beijing100144,China;2.PowerResearchInstitute,StateGridJiangsuElectricPowerCo.,Ltd.,Nanjing211103,JiangsuProvince,China)ABSTRACT:Anomalydetectiontechnologyhasanimportantengineeringpracticalsignificanceforbatterydatafeaturemining,retiredbatterycascadeutilizationscreeninggroupingandbatteryoperationstatesafetyevaluation.Therefore,thispaperproposesanewmethodofgeneticoptimizationanomalydetectionbasedontheclusteranalysisframe.Inthismethod,theclusteranalysisisfocusedonasacenterforanomalydetectionandtheswarmintelligenceoptimizationalgorithmisappliedasaneffectivewaytosolvetheglobaloptimizationproblem.Byreasonablydesigningobjectivefunctionstodescribedataanomaly,theeffectivedetectionofabnormaldataisrealized.Finally,takingtheabnormalstatedetectionofthebatterydataasanexample,bycomparingtheexistingmethodsandtheabnormaldetectionresultsunderthethreeclusteringideasproposedinthispaper,theadvantagesoftheproposedmethodinitspersonality,flexibilityandaccuracyofabnormaldetectionareverified,especiallyshowingabetterdetectioneffectintheclusteringoptimizationdetectionprocessbasedondensityidea,Itprovidesanewideaforreal-timebatteryabnormalstatedetectionanddatacleaning.KEYWORDS:batteryabnormaldetection;featureengineering;clusteranalysis;geneticoptimizationalgorithm摘要：异常检测技术对电池数据特征挖掘、退役电池梯次利用筛选分组以及电池运行状态安全评估均具有重要的工程实际意义。为此，该文提出一种基于聚类分析架构的...
