Intelligent Hybrid Wavelet Models for Short-Term Load Forecasting

《Intelligent Hybrid Wavelet Models for Short-Term Load Forecasting》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。

1、1266IEEETRANSACTIONSONPOWERSYSTEMS,VOL.25,NO.3,AUGUST2010IntelligentHybridWaveletModelsforShort-TermLoadForecastingAjayShekharPandey,DevenderSingh,andSunilKumarSinhaAbstract—Awaveletdecompositionbasedloadforecastap-Theelectricalloadatanyparticulartimeisusuallyassumedproachisproposedfor24-ha

2、nd168-haheadshort-termloadtobealinearcombinationofdifferentcomponents.Fromtheforecasting.Theproposedapproachisappliedtoandcomparedsignalanalysispointofview,loadcanalsobeconsideredasawithrepresentativeloadforecastingmethodssuchas:timese-linearcombinationofdifferentfrequencies.Thewavelettrans

3、-riesintraditionalapproachesandRBFneuralnetworkandformisintroducedtopreprocesstheloaddatainordertoen-neuro-fuzzyforecasterinnontraditionalapproaches.Theotherforecasters,suchasmultiplelinearregression(MLR),timeseries,hancetheaccuracyofforecasting.Inthiswork,inadditiontofeedforwardneuralnetwo

4、rk(FFNN),radialbasisfunctionneuralthehistoricalelectricitydemanddatathetemperature,whichhasnetwork(RBFNN),clustering,andfuzzyinferenceneuralnetworksignificantimpactonenergyconsumption,havealsobeenused.(FINN),reportedintheliteraturearealsocomparedwiththeTheloaddataaretransformedinlowandhighfr

5、equencycom-presentapproach.Theprocessoftheproposedwaveletdecompo-ponents.Itisshownthatthehighfrequencycomponentdoesnotsitionapproachisthatitfirstdecomposesthehistoricalloadandchangefromareferencedaytotheforecastdaywhereasthelowweathervariablesintoanapproximatepartassociatedwithlowfrequencies

6、andseveraldetailpartsassociatedwithhighfrequen-frequencycomponentissituation(climate)dependant.Henceaciescomponentsthroughthewavelettransform.Thehistoricalforecastmodelisdevelopedforthelowfrequencycomponentdataaresmoothenedbydeletingthehighfrequencycomponentsonly.andfedasinputtotheproposedm

7、odelsfortheprediction.Acom-AvarietyoftechniquesforSTLFbasedonconventionalsta-parisonofwaveletandnon-waveletbasedapproachesshowsthetisticalmethods,ANNbasedmodels,expertsystemsandhybridsuperiorityofproposedwaveletbasedapproachovernon-waveletmodelhavebeenpr