Page - 136 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 3283 cluster, anddormitorycluster, andcollectedtheirdailyelectrical loaddataoversixyears.Wedivided thecollecteddata intoa trainingset, avalidationset, andatest set. For the trainingset,weclassified electrical loaddatabypattern similarityusing thedecision tree technique. Weconsideredvarious configurationsforrandomforestandmultilayerperceptronandevaluatedtheirpredictionperformance byusingthevalidationset toselect theoptimalmodel. Basedonthiswork,weconstructedourhybrid dailyelectrical loadforecastmodelbyselectingmodelswithabetterpredictiveperformance insimilar timeseries. Finally,using the test set,wecomparedthedailyelectrical loadpredictionperformanceof ourhybridmodelandotherpopularmodels. Thecomparisonresults showthatourhybridmodel outperformsotherpopularmodels. Inconclusion,weshowedthatLSTMnetworksareeffective for reflectinganelectrical loaddependingon thedayof theweekand thedecision tree is effective in classifyingtimeseriesdatabysimilarity.Moreover,usingthese twoforecastingmodels inahybrid modelcancomplement theirweaknesses. In order to improve the accuracy of electrical load prediction, we plan to use a supervised learningmethodreflectingvariousstatisticallysignificantdata.Also,wewillanalyze theprediction performanceindifferentlook-aheadpoints(fromthenextdaytoaweek)usingprobabilisticforecasting. AuthorContributions: J.M.designedthealgorithm,performedthesimulations,andpreparedthemanuscript as thefirstauthor.Y.K.analyzedthedataandvisualizedtheexperimental results.M.S. collectedthedata,and developedandwrote the loadforecastingbasedontheLSTMnetworkspart. E.H.conceivedandsupervisedthe work.Allauthorsdiscussedthesimulationresultsandapprovedthepublication. Funding:This researchwassupportedbyKoreaElectricPowerCorporation(Grantnumber:R18XA05)andthe BrainKorea21PlusProject in2018. Conflictsof Interest:Theauthorsdeclarenoconflictof interest. References 1. Lindley,D.Smartgrids: Theenergystorageproblem.Nat.News2010,463, 18–20. [CrossRef] [PubMed] 2. Erol-Kantarci,M.;Mouftah,H.T.Energy-efficient informationandcommunication infrastructures in the smartgrid:Asurveyoninteractionsandopenissues. IEEECommun.Surv.Tutor.2015,17, 179–197. [CrossRef] 3. Raza,M.Q.;Khosravi,A.Areviewonartificial intelligencebasedloaddemandforecastingtechniques for smartgridandbuildings.Renew. Sustain. EnergyRev. 2015,50, 1352–1372. [CrossRef] 4. Hernandez, L.; Baladron,C.; Aguiar, J.M.; Carro, B.; Sanchez-Esguevillas,A.J.; Lloret, J.;Massana, J.A surveyonelectricpowerdemandforecasting: Future trends insmartgrids,microgridsandsmartbuildings. IEEECommun. Surv. Tutor. 2014,16, 1460–1495. [CrossRef] 5. Kuo,P.-H.;Huang,C.-J.AHighPrecisionArtificialNeuralNetworksModel forShort-TermEnergyLoad Forecasting.Energies2018,11, 213. [CrossRef] 6. Ahmad,A.; Hassan,M.; Abdullah,M.; Rahman,H.; Hussin, F.; Abdullah,H.; Saidur, R.A reviewon applications ofANNandSVMfor building electrical energy consumption forecasting. Renew. Sustain. EnergyRev. 2014,33, 102–109. [CrossRef] 7. Vrablecová,P.;Ezzeddine,A.B.;Rozinajová,V.;Šárik,S.; Sangaiah,A.K.Smartgrid loadforecastingusing onlinesupportvector regression.Comput. Electr. Eng. 2017,65, 102–117. [CrossRef] 8. Hong,T.;Fan,S.Probabilisticelectric loadforecasting:Atutorial review. Int. J.Forecast. 2016,32, 914–938. [CrossRef] 9. Hahn,H.;Meyer-Nieberg,S.;Pickl,S.Electric loadforecastingmethods: Tools fordecisionmaking.Eur. J. Oper. Res. 2009,199, 902–907. [CrossRef] 10. Moon, J.; Park, J.;Hwang,E.; Jun,S.Forecastingpowerconsumption forhighereducational institutions basedonmachine learning. J.Supercomput. 2018,74, 3778–3800. [CrossRef] 11. Chung,M.H.;Rhee,E.K.Potentialopportunities forenergyconservation inexistingbuildingsonuniversity campus:Afieldsurvey inKorea.EnergyBuild. 2014,78, 176–182. [CrossRef] 12. Moon, J.;Kim,K.-H.;Kim,Y.;Hwang,E.AShort-TermElectricLoadForecastingSchemeUsing2-Stage PredictiveAnalytics. InProceedingsof theIEEEInternationalConferenceonBigDataandSmartComputing (BigComp),Shanghai,China,15–17 January2018;pp.219–226. 136