資源描述：
《A Data Mining Approach for Spatial Modeling in Small Area Load Forecast》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、516IEEETRANSACTIONSONPOWERSYSTEMS,VOL.17,NO.2,MAY2002ADataMiningApproachforSpatialModelinginSmallAreaLoadForecastHung-ChihWuandChan-NanLu,SeniorMember,IEEEAbstract—Inacompetitivepowermarket,locationsoffutureandfromwhichpredictionofthefuturespatialloa
2、dgrowthcanloadgrowthhavetobedescribedwithsufficientgeographicpreci-beobtained.siontopermitvalidmarketingstrategyandsitingoffutureT&DDomainexpertforspatialloadforecastishardtofind.Dataequipment.Smallarealoadforecastwhichprovidesinformationofmining(DM)
3、techniquewhichissuccessfulinmanyindustrialfutureelectricdemandthatincludesspatialandtemporalcharac-teristics,isusefulforT&Dandmarketplanning.Domainexpertsapplications,canbeusedinthispurposetoextractautomaticallyforspatialloadforecastrequirelongtermpr
4、acticingandaredif-avalidandusefulinformationfromlargedatabases.Ingeneral,ficulttofind.Inordertocapturethemeaningfulassociationsbe-theDMprocessincludesfivebasicsteps[4]–[7].tweenspatialdataandtheloadchanges,andtoprovideauseful1)DataSelection:Thisstepi
5、ncludesidentifyingthedatatotoolforspatialloadforecast,adataminingtechniquebasedona“KnowledgeDiscoveryinDatabase(KDD)”procedureispro-bemined,thenchoosingappropriateinputattributesandposedtodetermineautomaticallythepreferential“scores”oflandoutputinfor
6、mationtorepresentthetask.Effectiveimple-usechanges.Theproposedspatialmodelingapproachisanex-mentationanduseofthetoolsrequiressignificantexper-ploratorydataanalysis,tryingtodiscoverusefulpatternsinspatialtiseinextracting,manipulating,andanalyzingdataf
7、romdatathatarenotobvioustothedatauserandareusefulinthespa-alargedatawarehouse.tialloadforecast.2)DataFilteringandPreprocessing:InvolvedherearebasicIndexTerms—Datamining,fuzzymodel,knowledgediscoveryoperationssuchastheremovalofoutliers,collectingthein
8、database,spatialloadforecast.necessaryinformationtomodeloraccountfornoise,de-cidingonstrategiesforhandlingmissingdatafields,ac-I.INTRODUCTIONcountingforknownchanges,andappropriatenormaliza-tion.NADEREGULATEDenvironment,retailmarketpartici-3)DataConve
