Seite - 110 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 3442 3.4.1. TimeSeriesModeler—ExpertModel (SectorWise) In the expert time seriesmodeler it automatically assigns themodel best suitedbasedon the system’sexpertise. For the industrial, agriculturalanddomestic sectors ithasassignedBrown’smodel andit is foundtobe theappropriatemodel. For thecommercial, tractionandothers’ sector, theexpert model has assignedARIMA(0,1,0)model; ARIMA(2,1,0)model andARIMA(0,1,0) respectively, automaticallyas theappropriatemodelsas inTable6. Therespectivedegreesof freedomandother parametersareshowninTable7. Table6.SummaryofExpertmodel. ModelID ModelType Industry Brown Agriculture Brown Domestic Brown Commercial ARIMA(0,1,0) TractionRailways ARIMA(2,1,0) Others ARIMA(0,1,0) Table7.Summaryof themodel. Model Statistics Ljung-Box No. ofOutliers StationaryR2 Statistics DF Sig. Industry 0.281 3.802 17 1.00 0 Agriculture 0.080 49.040 17 0.000 0 Domestic 0.432 6.242 17 0.991 0 Commercial 1.102×10−15 10.125 18 0.928 0 Traction/Railways 0.331 15.057 17 591 0 Others 5.310×10−16 20.114 18 0.326 0 3.4.2.Holt’sModel-ExponentialSmoothingwithTrend SeveralmodelssuchasBrown’smodel,Holt’smodel,Expertmodelanddampedtrendmodel wereanalysed.Andtheanalysisof theHolt’smodel is shownintheTable8. Table8.SummaryofHolt’smodel. Model Statistics Ljung-Box No. ofOutliers StationaryR2 Statistics DF Sig. Industry 0.291 2.371 16 1.00 0 Agriculture 0.103 43.582 16 0.000 0 Domestic 0.447 6.536 16 0.981 0 Commercial 0.422 3.726 16 0.999 0 Traction/Railways 0.394 35.017 16 0.004 0 Others 0.434 19.379 16 0.250 0 TEC 0.069 14.250 16 0.580 0 3.4.3. TimeSeriesModeler (ExponentialSmoothing-Brown) Theanalysisof theBrownmodel is shownintheTable9. 110