基于数据挖掘的智能电表异常数据分析摘要随着自动化和信息技术的不断发展，在高级计量架构中通常会使用到智能电能表。智能电能表主要可实现电能计量、实时监测、网络通信、信息存储、信息交互以及自动控制等功能，并且可向使用者提供由于季节、时区或者节假日所导致形成的不同费率，实现精确的用电负荷曲线记录，可促进用户进行更加合理的供电规划。随着智能电能表现场运行年限的不断增加，功能的不断扩展，出现的故障类型也越来越多样复杂。随着国家电网公司用电信息采集系统的建成运行，主站会采集到大量的电能表运行数据，包括电量、电压、电流曲线和各类电能表异常事件等。本文首先综合智能电能表功能单元与硬件设计主要侧重于对其在使用过程中所遇到的不同种类的故障做出归纳总结，使用的方法为故障模式影响分析方法以各个功能单元为对象，逐一分析智能电能表可能出现的故障现象并进行归类。并根据现场实测的真实数据进行统计，确定现场出现故障频率较高的功能单元。最后通过可靠性预计，进一步确定薄弱功能单元的关键部件，为今后优化电能表设计、提高电能表运行可靠性提供依据。为分析得出电能表运行数据中所包含的随机、模糊信息及其之间关联性，需引入合适的数据分析方法，结合随机性、模糊性等方法原理进行定性和定量分析。本文创新性地结合聚类分析与云模型算法，对采集的故障数据进行分析处理，根据数据特征计算得出5个云模型，并结合电能表实际运行情况，赋予其5个不同的定性概念，并归纳出96个电能表故障信息诊断模型，能够快速地对电能表故障定性、处理，从而节省故障消缺的人力物力，提高工作效率。关键词：智能电能表故障分析数据挖掘云模型ABSTRACTWiththecontinuousdevelopmentofautomationandinformationtechnology,smartmetersareoftenusedinAMI.Thesmartmetermainlyrealizesfunctionssuchasenergymetering,real-timemonitoring,networkcommunication,informationstorage,informationexchange,andautomaticcontrol.Itcanprovideuserswithdifferentratesduetoseasons,timezones,orholidays.Itcanachievetheaccuratecurverecordsofreal-timeloadandpromoteuserstomakemorereasonablepowersupplyplanning.Withtheever-increasingfieldservicelifeofsmartenergymetersandthecontinuousexpansionoffunctions,thetypesoffailuresaremorediverseandcomplex.Themainstationwillcollectalargenumberofpowermeteroperationaldata,includingelectricity,voltage,currentcurveandvarioustypesofabnormalenergymetereventswiththecompletionandoperationoftheStateGridCorporation'spowerconsumptioninformationcollectionsystem.Inthispaper,thesmartmeterfunctionunitandhardwaredesignaremainlyfocusedonsummarizingthedifferentkindsoffaultsduringtheuseofthesmartmeter.Themethodusedisthefailuremodeimpactanalysismethod.Eachfunctionunitisusedasanobject,andthepossiblephenomenaofthesmartmetermaybeanalyzedonebyoneandclassified.Accordingtotheactualdatameasuredinthefield,statisticsaremadetodeterminethefunctionalunitwithhigherfailurefrequencyatthesite.Finally,throughthereliabilityprediction,thekeycomponentsoftheweakfunctionalunitarefurtherdetermined,providingthebasisforoptimizingthedesignofthesmartmeterandimprovingthereliabilityofthesmartmeterinthefuture.Inordertoobtaintherandomandfuzzyinformationcontainedintheoperatingdataoftheenergymeterandthecorrelationbetweenthem,anappropriatedataanalysismethodmustbeintroduced,andqualitativeandquantitativeanalysisshouldbecarriedbasedontheprincipleofrandomness,fuzziness,andsoon.Thispaperinnovativelycombinesclusteranalysisandcloudmodelalgorithmstoanalyzeandprocessthecollectedfaultdata,calculatesfivecloudmodelsbasedonthedatacharacteristics,andcombinedfivedifferentqualitativeconceptswiththeactualoperationofthesmartmeter.Thepapersumup96diagnosis...
