Page - 41 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1009 4.Conclusions ThispaperproposesanovelSVR-basedhybridelectric loadforecastingmodel,byhybridizing theseasonalmechanism, the tentchaoticmappingfunction,andtheCSalgorithmwithanSVRmodel, namelytheSSVRCCSmodel. Theexperimental results indicate that theproposedSSVRCCSmodel significantlyoutperformsother alternative compared forecastingmodels. Thispaper continues to overcomesomeinherentshortcomingsof theCSalgorithm,byactionssuchasenrichingthesearch spaceandthediversityofthepopulationbyusingthetentchaoticmappingfunctiontoavoidpremature convergenceproblemsandapplyingseasonalmechanismtoprovideusefuladjustmentscausedfrom seasonal/cycliceffectsof theemployeddataset. Eventually, theproposedSSVRCCSmodelachieves significantaccurate forecastingperformances. Thispaperconcludessomeimportantfindings. Firstly,byapplyingappropriatechaoticmapping functions it couldhelpempower thesearchvariables topossessergodicitycharacteristics, toenrich the searching space, then, determinewell appropriate parameter combinations of an SVRmodel, to eventually improve the forecasting accuracy. Therefore, any novel hybridizations of existed evolutionary algorithmswith other optimizationmethods ormechanismswhich could consider thoseactionsmentionedaboveduringmodelingprocessarealldeserving to takea trial toachieve more interesting results. Secondly, only hybridizingdifferent single evolutionary algorithmwith anSVRmodel could contributeminor forecastingaccuracy improvements. It ismoreworthwhile tohybridizedifferentnovel intelligent technologieswithsingleevolutionaryalgorithmstoachieve morehigh forecastingaccurate levels. This couldbean interesting future research tendency in the SVR-basedelectric loadforecastingfield. Acknowledgments: YongquanDong thanks the support from the project grants: NationalNatural Science FoundationofChina(No. 61100167),NaturalScienceFoundationof JiangsuProvince,China(No.BK2011204), andQingLanProject, theNationalTrainingProgramof InnovationandEntrepreneurshipforUndergraduates (No. 201710320058);ZichenZhangthanks thesupport fromtheprojectgrant: PostgraduateResearch&Practice InnovationProgramof JiangsuProvince (No. 2017YXJ214);Wei-ChiangHongthanks thesupport fromJiangsu DistinguishedProfessorProjectbyJiangsuProvincialDepartmentofEducation. AuthorContributions:YongquanDongandWei-ChiangHongconceived,designedtheexperiments,andwrote thepaper;ZichenZhangcollectedthedata,performedandanalyzedtheexperiments. Conflictsof Interest:Theauthorsdeclarenoconflictof interest. References 1. Wan,C.; Zhao, J.;Member, S.; Song,Y. Photovoltaic and solar power forecasting for smart grid energy management.CSEEJ.PowerEnergySyst. 2015,1, 38–46. [CrossRef] 2. Xiao,L.;Wang, J.;Hou,R.;Wu, J.Acombinedmodelbasedondatapre-analysisandweight coefficients optimizationforelectrical loadforecasting.Energy2015,82, 524–549. [CrossRef] 3. Bunn,D.W.;Farmer,E.D.Comparativemodels forelectrical loadforecasting. Int. J. Forecast. 1986,2, 241–242. 4. Fan, G.; Peng, L.-L.; Hong,W.-C.; Sun, F. Electric load forecasting by the SVRmodelwith differential empiricalmodedecompositionandautoregression.Neurocomputing2016,173, 958–970. [CrossRef] 5. Ju,F.-Y.;Hong,W.-C.ApplicationofseasonalSVRwithchaoticgravitational searchalgorithminelectricity forecasting.Appl.Math.Model. 2013,37, 9643–9651. [CrossRef] 6. Hussain,A.;Rahman,M.;Memon, J.A.Forecastingelectricityconsumption inPakistan: Thewayforward. EnergyPolicy2016,90, 73–80. [CrossRef] 7. Pappas, S.S.; Ekonomou, L.; Karampelas, P.; Karamousantas, D.C.; Katsikas, S.K.; Chatzarakis, G.E.; Skafidas,P.D.Electricitydemand load forecastingof theHellenicpower systemusinganARMAmodel. Electr. PowerSyst. Res. 2010,80, 256–264. [CrossRef] 8. Vu,D.H.;Muttaqi,K.M.;Agalgaonkar,A.P.Avariance inflation factor andbackwardeliminationbased robust regressionmodel for forecastingmonthlyelectricitydemandusingclimaticvariables.Appl. Energy 2015,140, 385–394. [CrossRef] 9. Dudek,G.Pattern-basedlocal linearregressionmodels forshort-termloadforecasting.Electr. PowerSyst.Res. 2016,130, 139–147. [CrossRef] 41