外文资料及其翻译

AWaveletBasedApproachforFastDetectionofInternalFaultinPowerTransformersThepowertransformerisoneofthemostexpensiveelementsofpowersystemanditsprotectionisanessentialpartoftheoverallsystemprotectionstrategy.Thedifferentialprotectionprovidesthebestprotectionforpowertransformer.Itsoperationprincipleisbasedonthispointthatthedifferentialcurrentduringaninternalfaultishigherthannormalcondition.But,alargetransientcurrent(inrushcurrent)cancausemal-operationofdifferentialrelays.Then,studiesfortheimprovementofthetransformerprotectionhavefocusedondiscriminationbetweeninternalshortcircuitfaultsandinrushcurrentsintransformers.Themagnetizinginrushcurrenthasalargesecondorderharmoniccomponentincomparisontointernalfaults.Therefore,sometransformerprotectionsystemsaredesignedtohaltoperatingduringtheinrushcurrentbysensingthislargesecondorderharmonic.Thesecondharmoniccomponentinthemagnetizinginrushcurrentstendtoberelativelysmallinmodernlargepowertransformersbecauseofimprovementsinthepowertransformercorematerials.Also,ithasbeenseenthatthefaultcurrentcancontainhighersecondorderharmonicsthantheinrushcurrentduetononlinearfaultresistance,CTsaturation.thedistributedcapacitanceinthetransmissionline,whichtransformerisconnectedto,orduetotheuseofextrahighvoltageundergroundcables.Variousmethodshavebeensuggestedforovercomingthisprotectionsystemmal-operation.Thispaperpresentsawaveletbasedmethodfordiscriminationamonginrushcurrent,internalshortcircuit,externalshortcircuitandenergizinganditisnotaffectedbyCTsaturationanditisabletodetectinternalfaultswhiletransformerenergization.UnlikeArtificialNeuralNetworkandFuzzylogicbasedalgorithms.Thisapproachisnotsystemdependent.Theoperatingtimeoftheschemeislessthan10ms.TheDaubechiesmotherwaveletisusedwithasamplerateof5kHz.Then,thedifferentialcurrentsofthethreephasesaredecomposedintotwodetailsandonlythesecondlevelwillbeconsideredbyusingdb5motherwavelet.DiscreteWaveletTransformThewavelettransformisapowerfultooltoextractinformationfromthenon-stationarysignalssimultaneouslyinbothtimeandfrequencydomains.Theabilityofthewavelettransformtofocusonshorttimeintervalsforhigh-frequencycomponentsandlongintervalsforlow-frequencycomponentsimprovestheanalysisoftransientphenomenasignals.Variouswaveletfunctions,suchasSymlet,MorlertandDaubechiesareusedtoanalyzedifferentpowersystemphenomena.Themotherwaveletmustbeselectedperformedbasedonitsapplicationandthefeaturesofsignal.whichshouldbeprocessed.Inthispaper,Daubechieswaveletisused.Therearethreetypesofwavelettransform.WhichareContinuousWaveletTransform(CWT).DiscreteWaveletTransform(DWT)andWaveletPacketTransform(WPT).DWTisderivedfromCWT.Assumethatx(t)isatomevariablesignal,thentheCWTisdeterminedby(1):dtttxCWT)()(),(21(1)Where,andaretranslatingandscalingparameters,respectively.Also,)(tisthewaveletfunctionand)(tisthecomplexconjugateof)(t.Waveletfunctionmustsatisfy(2)andshouldhavelimitedenergy:0)(dtt(2)Then,thediscretizedmotherwaveletisasfollows:)(1)(0000,mmmnmnbtt(3)Where,0a1and0b0andtheyarefixedrealvalues.Also,mandnarepositiveintegers.DWTisexpressedby(4):)()(),(,kkfnmfDWTknm(4)Where,)(,knmisthecomplexconjugateof)(,knm.In(4),themotherwaveletisdilatedandtranslateddiscretelybyselectingandb.m0andmnbb00(5)DWTcanbeeasilyandquicklyimplementedbycomplementarylowpassandhigh-passfilters.ProposedAlgorithmIntheproposedalgorithm,theDWTisappliedtothedifferentialcurrentsofthreephases.TheDaubechiesDb-5typewaveletisusedasthemotherwaveletandthesignalsaredecomposeduptothesecond-level.Then,thespectralenergyandstandarddeviationofthedecomposedsignalsinthend2levelarecalculated.Theproposedmethodconsistsoftwosteps;detectionanddiscrimination.DisturbanceDetectionUndernormalconditionsandexternalfaults,thedifferentialcurrentshavesmallervaluesthaninternalfaults.Howeverinsomeoperatingconditions,theexternalfaultscanresultinhighdifferentialcurrentsduetoratiomismatchofCTsortapchangesofpowertransformer.Then,theseconditionsmaycausemal-operationoftherelay.Therefore,athresholdcurrentisusedinordertopreventmalfunctionscausedbynon-faultycurrents.Ifoneofdifferentialcurrentsexceedsthisthresholdvalue,itwillbeidentifiedasafault.Thethresholdvalueiddefined,asfollows:2)(sec.detCTperCTiiki(6)WhereCTisecandCTperiarethesecondaryandprimaryCTcurrents,respectively,andkistheslopeofthedifferentialrelaycharacteristic.Ifdifiidet,thenthedetectionalgorithmdefinesitasaninternalfault.DisturbanceDiscriminationInordertoclassifydisturbances,thedifferentialcurrentsaredecomposeduptothesecondlevel,usingDaubechiesDb5typewaveletwithdatawindowlessthanthehalfofthepowerfrequencycycle.Asamplingrateof5kHz,isconsideredforthealgorithm(i.e..100samplesperpowerfrequencycyclebasedon50Hz).Then,theenergyandstandarddeviationintheseconddetailarecalculatedforeachdifferentialcurrent.Itisseenthatthespectralenergyaswellasthestandarddeviationinnd2leveltendstohavehighvaluesduringinrushcurrents.Then,adiscriminationindex(mdD)canbecalculatedbymultiplyingthespectralenergybystandarddeviationintheseconddetailforeachdiffere