Comparison of correlation analysis and DTW used in distribution network topology verification

《Comparison of correlation analysis and DTW used in distribution network topology verification》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

2018ChinaInternationalConferenceonElectricityDistributionTianjin,17-19Sep.2018ComparisonofcorrelationanalysisandMSDusedindistributionnetworktopologyverificationZeyangTang,Member,IEEE,KunpengZhou,KanCao,DefuCai,XiaodongYu,LiWan,YuzeRaoandJunhuiXinproblemofverifyingandupdatingDNTinformationintheirAbstract—Manypowerutilitieshaveproblemswiththerecords.qualityofdataaboutdistributionnetworktopology.ThisThetraditionalmanualwayofverifyingDNTisaffectstheoperationandmaintenanceofsmartgrid,includingtime-consumingandlabor-intensive.Theoperationdataoutagemanagementandlineloss.Theoperationdata,suchasproducedbydistributionnetworkcannotonlydirectlyreflectvoltagedatawhosesimilarityhasbeenanalyzedtoverifytheoperationstateofdistributionnetworkbutcanalsodistributionnetworkconnectivity.Themostcommonlymethodindirectlyreflectthedistributionnetworktopology.Therefore,usedtoevaluatesimilarityiscorrelationanalysis.Inthispaper,researchershaveproposeddata-drivenmethodstosolvethecorrelationanalysisandmorphologysimilaritydistance(MSD)problem.W.Luan[5-8]hasproposedamethodofusingthehavebeencomparedwhenusedindistributionnetworkmeasurementdatatoverifythetopologyofthedistributionconnectivityverification.Theresultsshowthatthedatashouldnetwork.Inordertoverifythetopology,thesingle-phasebenormalizedwhenMSDisused.Fourcasestudieshavebeenvoltageisusedtocalculatethesimilarity.J.Geng[9-10]hascarriedoutinordertocomparetheperformanceofthetwoproposedthemethodbasedondiscreteFréchetdistanceandmethods.Resultsshowthatwhenthevoltagelossof10kVclippingneighbormethod,whichisusedtocheckthetopologyfeederislarge,theperformanceofcorrelationanalysisisbetteroflowvoltagedistributionnetwork.ThistypeofdistributionthanMSD.Whenthevarietyrangeofthevoltagecurvesisnetworktopologyverificationismainlyusedtoverifywhichsmall,theperformanceofMSDisbetterthancorrelationsubstationareathelow-voltageusersbelongto.L.Chen[11-12]analysis.Whenthevoltagecurveshavecatastrophepoints,thehasproposedtopologyverificationtechnologybyusingtheperformanceofMSDisbetterthancorrelationanalysis.automaticmappingofthedistributionnetworkmodel.ThistypeoftopologyverificationismainlyusedtoverifytheIndexTerms—distributionnetwork;topology;correlationrelationshipofthe10kVlinesofdistributionnetwork.analysis;morphologysimilaritydistanceHowever,thetransformeroutletvoltageisunbalancedwhenthethree-phaseloadisunbalanced.Thewrongresultswillbeobtainedifsinglephasevoltageisusedtocalculatesimilarity.I.INTRODUCTIONInordertoovercometheinfluencebringsbyunbalancedDistributionnetworktopology(DNT),asanimportantpartthree-phaseload,Z.Tang[13-14]hasproposedthereductionofbasicnetworkdata,hasrecentlyreceivedincreasingmethodofoutletvoltageoftransformerwithunbalancedattention.ThecorrectnessofDNTwillsignificantlyaffectthethree-phaseload.Equivalentbalancedoutletvoltagecurvecancalculationofdistributionnetworklossesandmaintenanceofbeobtainedbyusingthismethod.distributionnetwork.InordertoreduceactivenetworklossesAsintheresearchdiscussedabove,correlationanalysisis[1],tobalanceloadamongthethreephases[2-3],andtomostlyoftenusedtocalculatethesimilarityofthevoltageimprovepowersystemreliability[4],itisnecessarytoadjustcurvesinordertoverifyDNT.Inthispaper,wemainlyfocustheDNT.However,manypowerutilitiesarefacingtheontheusageofcorrelationanalysisandmorphologysimilaritydistance(MSD)inDNTverification.FourCasestudieshaveZ.Tang,K.Zhou,K.Cao,D.Cai,X.Yu,L.Wan,Y.Rao,J.Xinarewithbeenmadetocomparetheperformanceofthetwomethods.ElectricPowerResearchInstitute,StateGridHubeiElectricPowerCo.,Ltd.,Wuhan430077,China(whu_tzy@126.com;zhoukp2@hb.sgcc.com.cn;caok@hb.sgcc.com.cn;caidf4@hb.sgcc.com.cn;yuxd13@hb.sgcc.com.cn;wanl18@hb.sgcc.com.cn;raoyz1@hb.sgcc.com.cn;xinjh@hb.sgcc.com.cn;).CICED2018PaperNo.201805280000094Page1/41691

12018ChinaInternationalConferenceonElectricityDistributionTianjin,17-19Sep.2018II.METHODamethodforcalculatingtheequivalentbalancedtransformeroutletvoltageswhenthree-phaseloadsareunbalancedisA.Correlationanalysisproposedinthispaper.ThismethodcorrectstheunbalancedInordertoidentifytheerrorofadistributionnetworkthree-phasevoltagestotheirequivalentbalancedthree-phasetopology,thecorrelationcoefficientofdifferentdistributionvoltage.transformersoutletvoltagesshouldbecalculated.ThegreaterT1245T2theabsolutevalueofcorrelationcoefficient,thegreateristheT3probabilityofthetwodistributiontransformersbelongingto240thesame10kVfeeder.Accordingly,thevalueofthecorrelation235/Vcoefficientcanbecategorizedasfollows:0.8

22018ChinaInternationalConferenceonElectricityDistributionTianjin,17-19Sep.2018TheresultsofcorrelationanalysisandMSDafterdata10.5normalizationareshowninTABLEII.Itcanbefoundthatdata0.50.40.3normalizationdoesnotaffecttheresultsofcorrelation00.2coefficientandcanhelptoobtainthecorrectMSD.Inthe-0.50.1followingsections,thedatanormalizationwillbecarriedout-10beforecorrelationanalysisandMSDareused.TABLEIIRESULTSOFCORRELATIONANALYSISANDMSDAFTERDATANORMALIZATIONCorrelationcoefficientMSDTransformerT1T2T3T1T2T3(a)correlationanalysis(b)MSDT11.000.980.98-0.0330.036Fig.3.ResultsoftransformerT6andT7usingcorrelationanalysisandMSDT20.981.000.980.033-0.03BasedontheresultsofFig.2andFig.3,thethresholdvalueT30.980.981.000.0360.03-ofcorrelationcoefficientandMSDaresetas0.85and0.1.Whenthecorrelationcoefficientoftwotransformersisbiggerthan0.85ortheMSDofthetwotransformersislessthan0.1,itIV.CASESTUDYcanbeconcludedthatthetwotransformersareconnectedtheA.Case1same10kVfeeder.InordertoevaluatetheperformanceofthecorrelationB.Case2analysisandMSD,thedataoftransformerT4andT5fromththAsshowninFig.4,transformerT8andT9areconnectedtoDecember282016toFebruary222018hasbeencollected.thesame10kVfeederwithtransformerT8attheheadoftheTransformerT4andT5areconnectedtothesame10kVfeeder.thlineandtransformerT9attheendoftheline.ThevoltagelossThedataoftransformerT6andT7fromDecember102016tothisobvious.ThecorrelationcoefficientofthetwotransformersFebruary222018hasbeencollected.TransformerT6andT7is0.98whichmeansthatthetwotransformersareconnectedtoareconnectedtothedifferent10kVfeeders.thesame10kVfeeder.TheMSDofthetwotransformersis0.12.FromtheresultsofFig.2,itcanbefoundthatiftwoItcanbeconcludedthatwhenthevoltagelossof10kVfeederistransformersareconnectedtothesame10kVfeeder,thelarge,theperformanceofcorrelationanalysisisbetterthancorrelationcoefficientofthetwotransformersvoltagecurvesMSD.areverycloseto1.Theaverage,minimumandmaximum260T8correlationcoefficientoftransformerT4andT5are0.97,0.786T9and0.997respectively.Atthesametime,theaverage,255minimumandmaximumMSDoftransformerT4andT5are/V0.044,0.014and0.177respectively.eg25010Valto10.2V2450.80.155V0.60.12400.40.050.2235003691215182124Time/hourFig.4.EquivalentbalancedoutletvoltagecurvesoftransformerT8andT9C.Case3(a)correlationanalysis(b)MSDAsshowninFig.5,transformerT10andT11areconnectedtoFig.2.ResultsoftransformerT4andT5usingcorrelationanalysisandMSDthesame10kVfeeder.Theaverage,minimumandmaximumFromtheresultsofFig.3,itcanbefoundthatiftwovoltagevaluesoftransformerT10is232.5,228.8and234.9transformersareconnectedtothedifferent10kVfeeders,therespectively.ThevarietyrangeofthevoltagecurvesofcorrelationcoefficientofthetwotransformersvoltagecurvestransformerT10issmall.Thecorrelationcoefficientofthetwodiffers.Theaverage,minimumandmaximumcorrelationtransformersis0.82whichmeansthatthetwotransformersarecoefficientoftransformerT6andT7are0.542,-0.658andconnectedtothedifferent10kVfeeder.TheMSDofthetwo0.994respectively.Atthesametime,theaverage,minimumtransformersis0.05whichmeansthatthetwotransformersareandmaximumMSDoftransformerT6andT7are0.135,0.018connectedtothesame10kVfeeder.Fromtheaboveanalysis,itand0.451respectively.canbefoundwhenthevarietyrangeofthevoltagecurvesisCICED2018PaperNo.201805280000094Page3/41693

32018ChinaInternationalConferenceonElectricityDistributionTianjin,17-19Sep.2018small,theperformanceofMSDisbetterthancorrelationACKNOWLEDGMENTanalysis.ThispaperissupportedbyprojectofStateGridofHubei245T10T11powersupplycompany,No52153216001H.240REFERENCES/Veg[1]D.Hung,N.Mithulananthan,andR.Bansal,“IntegrationofPVandBESalt235unitsincommercialdistributionsystemsconsideringenergylossandoVvoltagestability,”AppliedEnergy,2014,vol.113,no.6,pp.1162-1170.[2]C.Wang,Y.Zhou,J.Wang,etal,“AnovelTraversal-and-Pruning230algorithmforhouseholdloadscheduling,”AppliedEnergy,2014,vol.102,no.2,pp.1430-1438.225[3]F.Guarino,P.Cassarà,SLongo,etal,“Loadmatchoptimisationofa3691215182124residentialbuildingcasestudy:Across-entropybasedelectricitystorageTime/hoursizingalgorithm,”AppliedEnergy,2015,vol.154,pp.380-391.[4]P.Zhang,W.Li,S.Li,etal,“ReliabilityassessmentofphotovoltaicFig.5.EquivalentbalancedoutletvoltagecurvesoftransformerT10andT11powersystems:Reviewofcurrentstatusandfutureperspectives,”AppliedEnergy,2013,vol.104,no.4,pp.822-833.D.Case4[5]W.Luan,Y.Yu,B.Wang,“AMIdataanalytics,”ProceedingsoftheCSEE,2015,vol.35,no.1,pp.29-36．260T12[6]W.Luan,J.Peng,M.Maras,etal,“Smartmeterdataanalyticsfor255T13distributionnetworkconnectivityverification,”IEEETransactionson250SmartGrid,2015,vol.6,no.4,pp.1964-1971.245[7]W.Luan,J.Peng,M.Maras,etal,“Distributionnetworktopologyerror/Ve240correctionusingsmartmeterdataanalytics,”PowerandEnergySocietygalt235GeneralMeeting,IEEE,2013,pp.1-5.oV230[8]D.Xu,W.Luan,PWang,etal,“Analysisandapplicationsofsmartmeter225data,”PowerSupplyandUtilization,2015,vol.32,no.8,pp.25-30．220[9]J.Geng,B.Wu,DWan,etal,“Low-voltagetransformerareatopology215verifcationbasedonoutlierdetection,”ElectricPowerICT,2017,no.5,2103691215182124pp.61-65.Time/hour[10]J.Geng,X.Zhang,Z.Guo,etal,“Topologyverificationoflow-voltagetransformerareasbasedondiscreteFréchetdistanceandeditingFig.6.EquivalentbalancedoutletvoltagecurvesoftransformerT12andT13nearest-neighborsmethod,”ElectricalMeasurement&Instrumentation,2017,vol.54,no.5,pp.50-55.AsshowninFig.6,transformerT12andT13areconnectedto[11]L.Chen,B.Han,J.Zhao,etal,“Automaticgraphingbasedverificationthedifferent10kVfeeders.Itcanbefoundthatthevoltagetechniquesofmodeltopologyofdistributionnetworkanditscurvesofthetwotransformershavehighsimilarityinsomepartimplementation,”AutomationofElectricPowerSystems,2017,vol.41,no.2,pp.160-164.whileatthetimeof15:00thevoltagecurvesofthetwo[12]L.Chen,Y.Zhao,T.Han,etal,“Hierarchicallystructuredandmodeltransformershaveobviousdifferences.Thecorrelationdrivenautomaticgenerationofdistributionnetworkgraphs,”Automationcoefficientofthetwotransformersis0.89whichmeansthattheofelectricpowersystems,2015,vol.39,no.1,pp.226-232.twotransformersareconnectedtothesame10kVfeeder.The[13]Z.Tang,K.Zhou,K.Caoetal,“SubstationareatopologyverificationMSDofthetwotransformersis0.29whichmeansthatthetwomethodbasedondistributionnetworkoperationdata,”HighVoltageEngineering,2018,vol.44,no.4,pp.1059-1068.transformersareconnectedtothedifferent10kVfeeder.From[14]Z.Tang,K.Cao,D.Cai,etal,“Reductionmethodofoutletvoltageoftheaboveanalysis,itcanbeconcludedwhenthevoltagecurvestransformerwithunbalancedthree-phaseload,”EnergyInternetandhavecatastrophepoints,theperformanceofMSDisbetterthanEnergySystemIntegration,IEEE,2018,pp.1-6.correlationanalysis[15]L.Yan,“Applicationofcorrelationcoefficientandbiasedcorrelationcoefficientinrelatedanalysis”JournalofYunnanUniversityofFinanceV.CONCLUSIONandEconomics,2003,vol.19,no.3,pp.78-80.[16]Z.Li,Y.Liu,T.Zhang,“TimeseriessimilaritymeasurementbasedonFourcasestudieshavebeencarriedoutinordertocomparemorphologysimilaritydistance,”Computeengineeringanddesign,2016,theperformanceofthecorrelationanalysisandMSD.Thevol.37,no.3,pp.679-683.resultsshowthatthedatashouldbenormalizedwhenMSDisZeyangTang(M’15)wasborninJingzhou,HubeiProvince,China,in1987.used.Resultsalsoshowthatwhenthevoltagelossof10kVHereceivedaB.Sc.degreeinElectricalEngineeringfromTianjinUniversity,feederislarge,theperformanceofcorrelationanalysisisbetterTianjin,China,in2009andaPh.D.degreeinElectricalEngineeringfromWuhanUniversity,Wuhan,China,in2015.HisresearchmainlyfocusesonthanMSD.Whenthevarietyrangeofthevoltagecurvesisdataanalysisandelectricaldistributionnetworks.small,theperformanceofMSDisbetterthancorrelationanalysis.Whenthevoltagecurveshavecatastrophepoints,theperformanceofMSDisbetterthancorrelationanalysis.CICED2018PaperNo.201805280000094Page4/41694