Seite - 113 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 3442 3.7. CurveEstimation 3.7.1. LinearModel ThisTable12presents theregressioncoefficientsandit is tobemadeanoteof that, thecorrelation will be of negative valuewhen the slope is negative. The following linear regression equation is determinedbythesecoefficients. y=−31,615,881.66+16,010.77x Table12.SummaryofLinearmodel. Equation Summaryof theModel ParameterEstimates R2 df1 df2 Sig. Constant Linear 0.844 1 42 0.000 −31,615,881.66 Series1 in theFigure11 is theactualTECandtheseries2 is the linear forecast. The forecasted value for the linearcurvefittingmodel for2030 is885,981.44MW. ͲϮϬϬ͕ϬϬϬ Ϭ ϮϬϬ͕ϬϬϬ ϰϬϬ͕ϬϬϬ ϲϬϬ͕ϬϬϬ ϴϬϬ͕ϬϬϬ ϭ͕ϬϬϬ͕ϬϬϬ ^ĞƌŝĞƐϭ ^ĞƌŝĞƐϮ Figure11.Chart forLinearmethod(1971–2030). 3.7.2.Compound/ExponentialModel The Table 13 represents the regression coefficients and it is to be taken into account that, the correlationwill be in the negative sidewhen the slope is of negative value. The following regressionequation ismadeoutof thesecoefficients. y=41,116.428e0.07x Table13.Summaryof themodel. Equation Summaryof theModel ParameterEstimates R2 df1 df2 Sig. Constant Comp./Exp. 0.991 1 42 0.000 41,116.428 113