基于广域测量的电力系统在线安全分析若干关键问题的研究

华中科技大学博士学位论文基于广域测量的电力系统在线安全分析若干关键问题的研究姓名：李大虎申请学位级别：博士专业：电力系统及其自动化指导教师：曹一家20060927II//PMU/SCADAPMUPMU/PMUIIIProny/IVABSTRACTTheappearanceofsynchronizedphasormeasurementbasedwide-areameasurementsystem(WAMS)whichisoneofthenewfrontsubjectsinpowersystemcanmakethedistributedsynchronizedphasormeasurement,dynamicsecuritymonitoring,onlinesecurityanalysisandwide-areastabilitycontrolpossibleforlargeinterconnectedpowergrid.Firstly,theconceptofwide-areameasurementsystemandapplicationsofthewide-areameasurementsysteminpowersystemareintroducedinthisdissertation.Secondly,thispaperhasstudiedthewide-areameasurementbasedreal-timedynamicmonitoring/stateestimation,loadmodelinganddynamicsecurityanalysismethods.Atlast,thisdissertationsummarizesalltheworkwhichhasbeendoneinthisdissertationandalsoproposesthepotentialprospectofnewresearchdirectionsofwide-areameasurementsysteminpowersystem.Thedissertationconsistsofsevenmainchaptersasfollows:Thefirstchapteranalyzestheshortcomingsofthetraditionalinformationacquisitionsystemandtraditionaldynamicsecurityanalysismethods,pointsouttheimportanceofthecredibleinformation,presentstwoconceptsofwide-areameasurementsystemandintroducesvariousresearchesandapplicationsofwide-areameasurementsysteminpowersystem.Atendofthispart,theconceptsandcorrespondingintroductionsofwide-areaprotectionandstrategypowerinfrastructuredefensesystemarealsogivenout.Thesecondchapteranalyzestheoptimalplacementofphasormeasurementunit(PMU).Duetothefactorsofthepriceandtechnology,nowitisimpossibletoplacePMUatallthebusesinthepowersystem.HowtousetheminimalnumberofPMUtoobtainthemaximalobservabilitytopowergridisthechiefproblemneedtosolveinthecurrentwide-areameasurementsystem.Inthispart,thedissertationhasproposedtwooptimalplacementmethodswhichcanmeettherequirementsofobservablityofpowergridorcancatchthetransientperformanceofallthegenerations,respectively.VThethirdchapterhasproposedahybridmeasurementsbaseddynamicstateestimationmethod.Dynamicmonitoringisthebasisofonlinesecurityanalysis,butthetraditionalsupervisorycontrolanddataacquisition(SCADA)systemcannotsupervisethepowersysteminrealtime.TheappearanceofPMUmakesthedynamicmonitoringbecomepossible.ButbecausetheplacementofPMUofcurrentstagemaynotguaranteethepowergridtobeobserved,thehybridmeasurementsbaseddynamicsupervisoryordynamicstateestimationbecomesthechiefdifficultyofcurrentresearch.Thisdissertationusestherelationshipofvoltagesensitivitiesofbuses’tosolvethisdifficulty.Thefourthchapterhasproposedawide-areameasurementbasedloadmodelingmethodbylocalmeasurementsandpseudo-measurementsofPMUs’.Bysupposingthegeneratormodelsandthetransmissionlinemodelsarerelativelyaccurate,theproposedmethodcanobtaintheallloadmodelsofeveryloadbusbyusingthetimedomainsimulationandevolutionaryalgorithm.Thefifthchapterhasproposedtwowide-areameasurementbasedtransientstabilityanalysismethods.Inthischapter,astabilityclassifierandapredictionmodelaresetupbyusingofdataminingtechnology,supportvectormachine(SVM)andsupportvectorregression(SVR)models.Theformerusesroughsettheoryandself-organizingfeaturemap(SOFM)toextractthefeaturesoftheinputsandpre-classifytheinputsamples,respectively,andatlasttheSVMcanbeusedtofindtheoptimalseparatinghyper-planeforthefuzzysamples,whichcanimprovetheprecisionoftheclassification.ThelattercanalsogethighpredictionprecisionbyusingSOFMtoclassifythesamplesandSVRtotrainthesamplesclassified.Thesixthchapterhasproposedawide-areameasurementsbasedvoltagestabilityanalysismethod.Thelocalmeasurementscanbeusedtoassessthevoltagestabilityproblem.Byusinglocalmeasurementsonelocalloadbranchcanbeseenasasimpleloadinfinitesystem,sowecanstudythelocalvoltagestabilityindexatfirst,andthenobtainthevoltagestabilityindexoftheallsystembysynthesizing.ThismethodcanmeettheVIrequirementofreal-timeanalysis.Butthetraditionalequivalentmethodcannotconsidertheeffectofthebranchcapacitanceandalsonotconsiderthechangeofloadcomponents.Inviewtothetwodefects,thisdissertationamelioratesthislocalequivalentsystembasedvoltagestabilityanalysismethod.Theseventhchapterhasproposedawide-areameasurementsbasedlow-frequencyoscillationanalysismethod.Pronyalgorithmisexcellentmethodtoidentifythelow-frequencymodes,butitissensitivetothenoise.Inthischapter,thedissertationcomparesseveralfiltermethodsandproposesanewnoisefilterfuzzylogicfilter,analyticalresultsindicatethatthefuzzyfilterisbetterthanothermethodsinnoisecancellation.KEYWORDS:Phasormeasurementunit(PMU);Wide-areameasurementsystem(WAMS);Optimalplacement;Dynamicsupervisory/dynamicstateestimation;Loadmodeling;Transientstability;Voltagestability;Low-frequencyoscillationI_____111.1[1-3][4][5,6][2,7][8-11]199678[12]1140020038145000300[13-15]20032[10,11,15]1.11.12020910kW[16]DynamicSecurityAssessment,DSA[17,18]31.1Table1.1Somerecentblackoutsinforeign19941214142001996721420019968109750200381461,800MW500029300199482420039286,400MW1996832143MW1580MW162003915Perlis,Perak,Kelantan,KedahPenang42005525150~200174120056223101..22005818141.2[19-88]1.2.1[19-37][16](DifferentialDifferenceAlgebraicEquations,DDAE)DDAE[19-23]