Page - 329 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1449 (4)Averageabsoluteerror (AAE) AAE= 1 n ( n ∑ i=1 |xˆi−xi|)/(1n n ∑ i=1 xi) (26) wherexand xˆare theactual loadandthe forecastedoneofchargingstation, respectively;nequals the numberofgroups in thedataset. Thesmaller theseevaluation indicatorsare, thehigher theprediction accuracy is. 4.3. ResultsAnalysis Theparametersof theproposedmodelaresetas: the totalwolfpackN=50, iterationnumber t=100, stepa=1.5, stepb=0.8,q=6,h=5. TheforecastingresultsareshowninFigure7. Figure7.Forecastingresultsof theproposedmodel. AscanbeseenfromFigure7, theproposedmodel isveryclose to theactual loadcurve ineach seasonandhasagooddegreeoffit. Figure8showstherelativeerrorof thepredictionresults. It canbe seenthat therelativeerrorof thepredictionresultsof theFC-WPA-LSSVMmodel is controlledwithin therange[−3%,3%],andthedegreeofdeviation isacceptable. Figure8.TheREof theproposedmodel. 329