外文翻译原文-SCADA电力系统的实时数据库管理概述

AbstractASupervisionControlandDataAcquisition(SCADA)systemisacommunicationandcontrolsystemusedformonitoring,operationandmaintenanceofenergyinfrastructuregrids.Comparedwithtraditionalapplications,aSCADAsystemhasaharshdeadlineforcriticaltasks.ThereisspecialtimeconstraintfortherealtimedatabaseusedinaSCADAsystem.TherealtimedatabaseinSCADAextendstraditionaldatabasetoincludein-memorydatabase.Suchrealtimedatabasemanagementaredesignedtooperateintheharshenvironmentofreal-timesystems,withstrictrequirementsforresourceutilization,andarereadytoprovidetheperformanceandreliabilityrequiredbyreal-lifeapplications.Inthispaper,themainprincipleofrealtimedatabasehasbeenintroduced.ItsimplementationinpowersystemSCADAisdiscussedandasampledatabaseisbrieflyintroduced.Keywords:SCADA,Real-time,Database,In-memoryDatabase1.IntroductionASupervisionControlandDataAcquisition(SCADA)systemisacommunicationandcontrolsystemusedformonitoring,operationandmaintenanceofenergyinfrastructuregrids.Comparedwithtraditionalapplications,aSCADAsystemhasaharshdeadlineforcriticaltasks.Ithasdifferentcharacteristics:thesystemmustmaintainmassivesharingdataandcontroldata;alsoeachreal-timetaskofaSCADAsystemhasarigoroustimerequirement,whilethedataforanalysisandprocessingistime-variant.Realtimedatahasashortlifecycle;forexample,allthedataofremotemeasurementsandremotesignalsmustbeupdatedevery5secondsandthedecisionorderivationbasedonobsoletedataisinvalid1,2.Therefore,thereisspecialtimeconstraintforthedatabaseusedinaSCADAsystem.Thedatabaseshouldbeabletoprocesspersistentstaticdata,maintaintheintegrityandconsistencyofthem,shouldbeabletodealwiththedynamicdatawithinitstimeconstraintsduringprocessingandensuretheconcurrenceandefficiencyofdataaccess3,4.Thiskinddatabaseisclassifiedasarealtimedatabase.Thedataanditsprocessingofareal-timeSCADAdatabasehaveharshtimeconstraints.Theaccuracynotonlyreliesonthebusinesslogicalresult,butalsoisdecidedbywhetherthelogicalderivationcompletesincertaininterval.Therefore,weneedtocombinereal-timetheorywithtraditionaldatabasetechnologytorealizeareal-timedatabaseinaSCADAsystem.2.TechnicalIssuesinReal-timeDatabaseImplementationLiketraditionaldatabasemanagementsystems,real-timedatabasesystems(RTDBMS)serveasrepositoriesfordataandprovideefficientstorage,loadandmanipulationofdata.TheymustbeequippedwithbasicfunctionsofgeneralDBMS.Moreover,tomeetthetemporalrequirementsofthemanageddata,timingconstraintsontransactions,andperformance,thereal-timedatabaseshouldsupporton-linequery,location,deletion,andinsertion.Anin-memorydatabaseisoftenusedtoachievethespeedrequirement.Functionalitiesofareal-timedatabasecanbesummarizedasbelow3,4:ProvidesplatformtosharedatawithentireEnergyManagementSystem(EMS)Providesopendatabaseinterface,realizesthefunctionofdatabasesuchasfastload,input,deletion,inquiry,load,andaccess;Providessnapshot,historydatastorage,copyandreutilization;Providesvaliditycheck;ProvidessecurityprotectionTorealizethefunctionalitiesofareal-timedatabase,thereareseveraltechnicalissuesinreal-timedatabaseimplementationaselaboratedbelow.2.1Real-timeDatabaseModelDesignandImplementationArelationaldatabaseisgoodatdatarepresentationandmanagement.WhenarelationaldatabaseisusedinaSCADAsystem,whichisalarge-scalecomplicatedreal-timesystemwiththefunctionsofdatamanagement,monitoring,analysisanddecision,andsomeshortcomingsOverviewofReal-TimeDatabaseManagementSystemDesignforPowerSystemSCADASystemJianWu,YongCheng,andNoelN.SchulzDepartmentofElectricalandComputerEngineeringMississippiStateU620-4244-0169-0/06/$20.002006IEEE.Authorizedlicenseduselimitedto:HuaiyinInstituteofTechnology.DownloadedonMarch10,2010at21:37:30ESTfromIEEEXplore.Restrictionsapply.areexposed.Forinstancesincedifferentvendersofpowersystemapplicationsdevelopedreal-timedatabaseindependently,thereisnostandarddatarepresentation,whichcausesintegrationproblems.ThecommonInformationModel(CIM)proposedbytheElectricPowerResearchInstitute(EPRI)canbeusedtoaddressthisissue5.2.2Real-timeTaskSchedulinganditsConcurrencyControlsInadditiontomaintainingdatabaseconsistencyasinconventionaldatabases,real-timedatabasesystemsmustalsohandletransactionswithintimeconstraints.Schedulingreal-timetransactionsisfarmorecomplexthantraditionalreal-timescheduling.Thesystemmustguaranteethemosttime-criticaltasktobeexecutedasearlyaspossible.Thatis,thesystemshouldbeabletoschedulethetransactionsexecutionsequenceaccordingtothepriority2.2.2DataStorageandSharingMemoryManagementTraditionaldiskdatabaseoperationisbasedondiskI/O.ButthediskI/Otimedelayanditsuncertaintysometimesisfataltothereal-timetask.Thereforeanotheressentialissueofreal-timedatabasesistoeliminatetimedelayanduncertainty.Thisneedsthesupportofdatabaseinmemory.Toensurereal-timetasksaccessthedatabasefastandaccurately,twoproblemsshouldbesolved6.(1)Guaranteesallthekeytasksofloadandstorehappeninmemoryonly.(2)Realizesdatasharingamongtasks.2.3DatabaseSecurityandRecoveryIfthedatabasesystemcrashes,atraditionaldatabasesystemusesalogtorestorethedatabasetowhatitwasatthetimeofthecrash,butreal-timedatabaserestorationismorecomplexduetothefactsbelow3:(1)Theprocessofrestorationblocksthereal-timetaskforaccessingthedatabaseandcausesreal-timetaskovertime.Itispossibleforactivereal-timetasktoaccesstransientdataandmakethewrongdecision;(2)Inthereal-timedatabasemuchdataistemporaryandvolatile.Sometimestheinfluenceofdatainconsistencyandincorrectistemporaryandnon-continuous.Forexample,incorrectremotemeasurementwillaffectthestateestimationhappenedatsametimeintervalonlybutnotafterthisinterval.Therefore,therecovertimeofdifferenttypesofcrasheddataisdependentupontheapplicationthatusedthedata.3.DataObjectofDispatchSystemClassificationBeforebuildingadataschemainrealtimedatabaseitsnecessarytoknowthedataobjectcharacteristic.ThedatainSCADAsystemcanbedividedintothreetypesofdataobjects:RealtimeDataObject,StaticDataObjectandDerivedDataObject1,2,3.3.1Real-timeDataObjectReal-timedatasuchasremotemeasurementandremotesignalreflectsthestatusofapowersystem.Thecommoncharacteristicofthedataisthestricttimeconstraint.Thedataiswrittenintothereal-timedatabaseperiodically.Areal-timedataobjectisassociatedwithatimestampandlifecycle.Theyareonlyvalidfortherespondingsamplingtime.ThetaskofquerythesamplefromRemoteTerminalUnit(RTU)anddatabasewrittenismanagedbyreal-timetaskdispatchmanagementasstatedinnextsection.Oncethevalueofrealtimedataobjectiswrittenintothedatabase,itcannotbechangedbyothertask.Newsampledatawillbestoredinareal-timedatabaseasanewdataobject5.3.2StaticDataObjectAstaticdataobjectisconstant.Itisaspecialdatatypeinreal-timedatabases,anditsvaluedoesnotchangewithtime.Thetimestampassociatedwiththedataissystemcreatingtimeortherenewaltime.Thistypeofdata,suchasparametersoftransmissionlineandtransformerimpedance,isspreadwideinSCADAsystems.3.3DerivedDataObjectAderiveddataobjectiscalculatedfromagroupofrealtimeDataObjectsandotherdataobjects.Therefore,thederiveddataobjectalsohasatimestampandlifecycle.Thetimestampofaderiveddataobjectisthederivationtime.Thelifecycleofthistypedataisthetimeintervalbetweenthisderivationtimeandthenextderivationtime.Thevalueofaderiveddataobjectinadatabasemayberenewed.Itsvaluemaybepreservedornotdependingonitsfunctionality.Thistypeofdataincludespowergridnetworkmodel,magnitudeandangleofbusvoltage,andbranchcurrentobtainedfromstateestimationinEMS.Thedatacouldbethesolutionfromonemodule,ortherawdataofsometask.4.Real-timeDatabaseSystemStructure63Authorizedlicenseduselimitedto:HuaiyinInstituteofTechnology.DownloadedonMarch10,2010at21:37:30ESTfromIEEEXplore.Restrictionsapply.Inordertosupportthefunctionalitiesabove,arealtimedatabasecanbeanintegrationoftraditionalrelationaldatabaseandin-memorydatabase.ThegeneralarchitectureisshowninFigure11,2.Itincludesthereal-timetaskdispatchmanagement,in-memorydatabase,I/Odispatchandtherelationaldatabase.Real-timetaskdispatchmanagementReal-timeapplicationIn-memorydatabaseDatabasemodelmanagementReal-timeresourcemanagementDataoperationNetworkcommunicationI/ODispatchRelationalDatabaseFigure1.RealtimedatabasesystemstructureThereal-timetaskdispatchmanagementprovidescoordinationbetweenactivitiesofreal-timetaskstoaccesstherealtimedatabase.Thein-memorydatabaseholdsthemaineffectiverangeofdatainthememorytoavoiddiskI/Oduringreal-timebusinessimplementationandenhancesimplementationefficiency.Itisthecorepartofthedatabase.Thedetailedelaborationwillbepresentedinsection5.Therelationaldatabaseinthisarchitecturecanbeusedasinterfaceforfurtherdevelopment,aswellasstoragemediumforin-memorydatabase.TheI/Odispatchisresponsiblefordatasynchronizationofmemorydatabaseandrelationaldatabase,throughwhichthereal-timedatabasecanbeseamlesslyintegratedwithtraditionaldatabase.4.1Real-timeTaskDispatchManagementInSCADA,therearenesting,communicationandcooperationrelationsamongreal-timetasks.Thederiveddataobjectofataskispossiblytheinputdataobjectofanothertask.Thedataobjectandthetaskaredependentoneachother.Itisnecessarytocoordinateactivitiesofreal-timetasks.Thetaskdispatchmanagementshouldprovidethefollowingfunctionstosharedatabaseresourcesandcoordinateconflict8:Startandstoptaskatfixedtime;Allowsetupconditionstotriggertask;Specifyexecutionsequenceamongtasks;Constraintheinstancenumberofglobaltask.4.2I/OdispatchTheI/Odispatchisresponsiblefordatasynchronizationofin-memorydatabasewithrelationaldatabase.Thein-memorydatabasehasaone-to-onerelationwithrelationaldatabase.Thedatainarelationaldatabaseneedstobeautomaticallysynchronizedwithin-memorydatabase.Itisimportanttodecidewhentowritethedatainanin-memorydatabasetoarelationaldatabaseandwhentoexportthedatainarelationaldatabasetoin-memorydatabase.Thereisnogeneralsolution.Onepossiblewayistoprovidemethodsofperiodicalupdateandforcedupdateindualdirectionbetweenarelationaldatabaseandanin-memorydatabase.Fordifferentapplications,thetwomethodscanbeusedfordifferentobjectsaccordingtotheircharacteristics.5.In-MemoryDatabaseAnin-memorydatabaseisthecorepartofareal-timedatabase.Itfunctionscoverthedatabasedatamodel,dataoperation,real-timeresourcemanagement,andnetworkcommunication.Thetraditionaldatabaseisadisk-baseddatabase.Itsprocessingtimeisindeterministic,becauseitinvolvesthediskaccess,datatransferbetweeninternalandexternalmemory,buffermanagement,waitinglistandlockmanagement.Thispropertymakestraditionaldatabasesystemunabletoachievetherequirementsofreal-timetransactionsthatareofhighefficiencyandtimedeterministic.Therefore,theintroductionofin-memorydatabasemakesitpossibletohavemosttransactionshappeninthememory,whichavoidsdiskI/Oduringreal-timetransactions,andreducesuncertaintyandenhancesefficiency6,8.5.1DataModelofDatabaseAnin-memorydatabase,astheextensionofarelationaldatabase,hassimilaritywitharelationaldatabase.Atableisthekeypartofanin-memorydatabase.Atabledescribesagroupdataobjectentityofthesametype.Whileatableistwo-dimensionalinrelationaldatabase,atableinanin-memorydatabasecanbemulti-dimensional.Althoughatwo-dimensionaltableisdeceptivelysimpletodesignandoperate,thissimplicityrequirescomplexjoinsofmany64Authorizedlicenseduselimitedto:HuaiyinInstituteofTechnology.DownloadedonMarch10,2010at21:37:30ESTfromIEEEXplore.Restrictionsapply.tableswhendoingdataanalysisoverhistoryoroveroneproperty5.Theintroductionofamulti-dimensionaltablestructureallowsanin-memorydatabasetoprocessdataoverhistoryoroveronepropertyandsavestoragespaceinmemory.Furthermoreitspeedsupusersaccesstodatafromanydimensionandfacilitatesdataanalysis.5.2Real-timeResourceManagement5.2.1ClassificationofDataStorageFormTherearetwotypesofdatastorageinadatabase,storageinin-memorydatabase,andstorageinthedisk.InaSCADAsystemtheamountofdataisextremelyhuge.Notalldatamustbeputintoanin-memorydatabase.Datastorageformcanbedecidedaccordingtothefollowingdatacharacteristics3:Timeliness:inareal-timedatabaseeachdataobjecthasalifecycle;thedatawithashortlifecyclemustbepreservedinanin-memorydatabase.Effectiveness:thedatawithfrequentaccessmustbestoredinthememorydatabase.Crucialnature:Inordertoguaranteethesystemefficiency,thecrucialdatashouldbepreservedinthein-memorydatabase.5.2.2DataSecurityandRestorationinDatabaseThereal-timetaskcannotbeinterruptedduringtherestorationprocess,whichmeansthatthedatarestorationshouldnotaffecttheoperationofareal-timesystem.Atraditionaldatabaseisusuallyrestoredbyusingrollbacksegment.Butitisdifficultforthismethodtosatisfythereal-timerequest.Asimplereliablerestorationmethodistocreateamirrordatabaseforessentialdata.Amirrordatabasecanbackuptheentiredatabaseatlocalorotherpointsofthenetwork.Whenthedataiscrashedatalocalsite,datacanberestoredfromthebackuppoint,whichenhancestheessentialdatasecurity7.5.3DatabaseOperation5.3.1FastAccessInterfaceFastdataaccessisbasicrequirementofareal-timedatabase.Inordertoenhanceoperatingefficiencyofsystemanalysisanddecisionsoftware,areal-timesystemneedstoprovideafastdatabaseaccessinterface.Throughthisinterface,databaseaccessefficiencycanbethesameasmemoryvariableoperation.Onepossiblerealizationistomapanentirepartitionofthedatabasetothesharedmemory.AndreturntheresultinaCstructurethroughanApplicationProgramInterface(API).Suchamechanismprovideseffectivedatamanipulationabilityandcircumambulatesdatabasesecurityanddataintegritycheck,whichintroducesacertaindegreeofrisk.Howevertheguidelineofthisinterfacedesignforareal-timedatabaseistogetpartiallycorrectbutpromptdata,ratherthanaccuratebutobsoletedata.Thus,suchriskistolerable1,StandardI/OTomakeupfortheinsufficiencyofafastaccessinterface,thedatabasecanprovideanotheraccessinterface-standardI/O.Thisaccessinterfaceneedstoprovidedatabasesecurityanddataintegritycheck.Italsoneedstoprovideaneffectivenetworkaccessmechanismandstrictconcurrencycontrol.Asthemanipulationofanydatainadatabasemustinteractwithadatabaseserver,thiskindofvisitmechanismhasreducedefficiencycomparedwithafastaccessinterface.Thiskindofmechanismcanbeusedinadistributedapplication,human-machineinterface(HMI),andinterfacewithotherapplications1,2.5.4CommunicationThebasicrequestsofreal-timedatabasenetworkcommunicationarehighefficiencyandreliability.Manymechanismssuchastheclient/serverinteraction,streamingtransmission,andcreationofamirrordatabaseatothersitescanbeusedtorealizedatadistribution.Thesemechanismshavedifferentadvantagesanddisadvantagesandareapplicabletodifferentapplicationarea7,8,Client/ServerInteractionIntheclient/serverinteractioncommunication,aclientaccessestheremotedatabase(server)throughthelocalreal-timedatabase.Foreachaccessthereisaseriesofinteractionsbetweentheclientandtheremoteservertoestablishtheconnection.Thisisatypicalapplicationofclient/serverusedforamajorityofdatabasemanagementsystems.Asthesizeofmessageissmallandinteractionfrequencyishigh,itiseasytocausenetworkoverburden,lowefficiencyandserveroverload.Theadvantageofclient/serverinteractionisnodatasynchronizationnecessaryamongdatabasesanditsapplicableforsmallamountofrealtimedatasharing.5.4.2StreamingTransmissionSimilartoclient/serverinteractions,streamingtransmissionsdeliverthedatatotheclientaccordingtorequest.Butthebasicprincipleofstreamingtransmissionisthattheserverwilldeliverthedatatotheclientperiodicallyasrequestedbytheclientatthefirstconnection.Thus,itcanavoidtheclient/serverfrequentconnectionestablishmentprocess.Also,theservercanreorganizemessagesaccordingtoclientrequesttoincreasetheaveragemessagesizeandreducetheframespassingthroughthenetwork.Thustransmissionefficiencyisenhanced.5.4.3CreationofMirrorDatabaseatOtherSiteAdifferentsitemirrordatabasecanhelptokeepdatasynchronizationefficiently.Itisapplicabletospecial65Authorizedlicenseduselimitedto:HuaiyinInstituteofTechnology.DownloadedonMarch10,2010at21:37:30ESTfromIEEEXplore.Restrictionsapply.applicationswhenmassivedataistransferred.Themainprincipleistomirrorsomebasicunitofthedatabase,suchasthewholedatabase,apartitionofthedatabaseortablestoothersites.ThistechniquehelpstoquickstartSCADAononenodewiththemirrordatabasespecifiedatothernode.Moreover,itcanhelptofastupdaterealizethedatabaseremotely.6.RealtimedatabaseimplementationSincedesignsofEMSdatabaseshaverevolvedaroundstringentreal-timeperformancerequirementsasstatedabove,specializedproprietarydatabasesareusedtoadapttohightransactionrates,highvolumesofinformation,andthediversityofdatacharacteristicoftheEMSenvironment.Atthesametime,tomakerealtimedatabaseplugcompatibleapplicationsoftware,standardinterfacearerequiredtoprovidedataaccess10.OneoftheleadingrealtimedatabasesisthePIsystem,developedbyOSIsoft.PIsystemisalargereal-time/historicaldatabaseusedfordatacollection,storageandmonitor.PIdatabaseusesswinging-doorcompressionandparticularfiltrationtechniquetoprocesstherawdatabeforeenteringthearchive,andachievesefficientarchivestoragewithoutlosingsignificantdata.PIdatabaseplatformalsoprovidesstandarddataaccessthroughitsDataAccessPackage(DAP)includingAPI,SDK,ODBCandOLEDB11.7.SummaryTherealtimedatabaseinSCADAextendsraditionaldatabasetoincludein-memorydatabase.SuchRTDBMSaredesignedtooperateintheharshenvironmentofreal-timesystems,withstrictrequirementsforresourceutilization,andarereadytoprovidetheperformanceandreliabilityrequiredbyreal-lifeapplications.Inimplementationofthereal-timedatabaseforadispatchautomationsystem,weneedtoconsidertheusestandarddatamodel-CIMforthedataobjects.Anin-memorydatabaseisthecorepartofareal-timedatabase.Specialmechanismsarerequiredforreal-timetaskschedulingandconcurrencecontrol,datastoragemanagementandcrashrecovery.Alsospeedinterfaceneedstobeprovidedtohandledataaccess.Aboveall,tillnowrealtimedatabaseisnotacommercialone,In-memorydatabaseneedtodesignaccordingtoapplication,weneedtoconsiderthemanagementsystemdesignandalsodatabaseschemadesign.Thispaperdiscussesthetechniqueissuestoimplementareal-timedatabaseandprovideageneralguidelineforthekeyfunctions.Finally,asamplerealtimedatabase,PIdatabaseplatformisbrieflyintroduced.8.AcknowledgementsThisresearchworkh