Seite - 351 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1282 Table10.Accuracyestimationof thepredictionpoint for the test set. Prediction Model <1% >1%and<2% >2%and<3% >3% Number Percentage Number Percentage Number Percentage Number Percentage BA-ELM 61 84.72% 10 13.89% 1 1.39% 0 0 ELM 33 45.83% 33 45.83% 6 8.34% 0 0 BPNN 24 33.33% 37 51.39% 10 14.29% 1 1.39% LSSVM 27 37.50% 26 36.11% 18 25% 1 1.39% Table11.Averageforecastingresultsof fourmodels. Index Model BA-ELM ELM BPNN LSSVM RMSE(MW) 5.89 11.08 12.74 14.47 MAPE(%) 0.49 1.13 1.29 1.43 MAE(MW) 4.27 9.81 11.14 12.51 Figure14.Root-mean-squareerror (RMSE)ofdifferentmodels in testingperiod. Figure15.Meanabsolutepercentageerror (MAPE)ofdifferentmodels in testingperiod. Figure16.Meanabsoluteerror (MAE)ofdifferentmodels in testingperiod. 351