Page - 114 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 3442 Theforecastedvalueforthecompoundcurvefittingmodel for2030is2,741,903.862MWisplotted inFigure12. Ϭ ϱϬϬ͕ϬϬϬ ϭ͕ϬϬϬ͕ϬϬϬ ϭ͕ϱϬϬ͕ϬϬϬ Ϯ͕ϬϬϬ͕ϬϬϬ Ϯ͕ϱϬϬ͕ϬϬϬ ϯ͕ϬϬϬ͕ϬϬϬ ϯ͕ϱϬϬ͕ϬϬϬ dŽƚĂů ůĞĐƚƌŝĐŝƚLJ ŽŶƐƵŵĞĚ &ŽƌĞĐĂƐƚ džƉŽŶĞŶƚŝĂů Figure12.Chart forExponentialmethod(1971–2030). Similarly theregressionequations forseveralothermodelsshallalsobe interpreted. 4.Discussion Themeasureoftheadequacyofthefit isdeterminedbythesamplecorrelation(r)betweenthetrue valueandresponsesgotoutof thefit. Thesamplecorrelation’s square isworkedout readilyoutof the statisticalpackage in theANOVAandis termedthecoefficientofdetermination(R2). Thecoefficient ofdetermination iscomputeddirectlybyestimatingPearson’scorrelation ‘r’betweenthepredicted andtheactualdata. Thecoefficientsofdeterminationaregenerallyexpressed in termsofpercentage. ThevalueofR2 lies inbetween0%and100%.Thenearer thevalue to theupperbound; thehealthier willbe thefit [35]. LEAPandHolt’sexponential smoothingmethodwerealsoemployedtoestimate theelectricity energydemandfor2030 inMaharashtra, India in thatstudy.ANN,multipleregressionapproaches andANOVAwereused. It isevident fromtheanalysisofvariance in thisarticle that theregression methodisable to forecast thecuttingforceswithahigheraccuracy[36]whichsupports thepresent study. Anoptimal renewable energymodel,OREMfor Indiawas evolved for theyear 2020–2021 tomeet the increasingenergyrequirements [37]. Anoptimizationmodel forvariousend-useswas formulatedbydeterminingtheoptimumallocationof renewableenergyfor2020–2021,byconsidering the energy requirement of the commercial sector. This study revealed that the social acceptance of bio resources increased by 3% and solar PV utilizationdecreased by 65% [38].Various energy demandforecastingmodelswerereviewedby[39]andfoundthat traditionalmethodsviz., timeseries regression,econometricanalysisareextensivelymadeuse fordemandsidemanagementwhereas the TECiscalculatedfor2030 in thispaper. Regressionanalysis, linearmodelanalysisandR2 correlation valuewasbuiltby [40] foracurvedvanedemisterwhichsupports theusing linearmodelanalysis of thecurrentstudy. Theutilizationofblackboxapproachto forecast theTECfor India issupported byvast literature amongwhich an optimal renewable energymodel for India for 2020–2021was presentedbydistributingrenewableenergyeffectively tohelp thepolicy framers inmarketing the renewableenergyresourcesandtodeterminetheoptimizedallotmentofvariousnon-conventional energy resources forvarious end-uses. In this study linear aswell asmultiple regressionanalysis provesGDPisagain thebetterpredictor in termsofMultiple linearregression. Therefore,asensible 114