Page - 327 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1449 (5) Themechanism of competition and regeneration of thewolf pack. In encirclement and suppression, thewolves that fail toget foodwillbeeliminatedandtherestofwolveswillberetained. Simultaneously,newwolvesarerandomlygenerated in thesamenumberas theeliminatedones. (6) Judgewhether themaximumnumberof iterationshasbeenreached. If themaximumnumber of iterationshas been reached, thepositionof thewolf is output; that is, the optimal value of the LSSVM’sparameters. If themaximumnumberof iterationshasnotbeenreached, thenreturntostep2. 3.4. Establishmentof theHybridForecastingModel Thispaperfirstlyanalyzes the influential loadfactors forquick-changee-buschargingstations, andFC is implemented toextract similardays to thepredictedoneas the training samples. Then, WPAis integratedwith theLSSVMmodel toobtain theoptimalvaluesofγandσ2. Finally,ananalysis isperformedonthe forecastingresults. The frameworkof theproposedhybridapproach isdisplayed inFigure6. Initialize the location of wolf pack and parameters of LSSVM Select the wolf at the location with the best target function as the head one The head wolf pushes wolf pack to update their positions through call to action and keep close to the prey The head wolf sends signals to wolf pack after finding the prey so as to complete the encirclement and suppression The wolves that fail to get food will be eliminated and randomly generate new wolves with the same number Whether the maximum number of iterations is reached? WPA Start Analyze the influential factors of load in e-bus charging station Collect data and implement FC Standardize the data Establish fuzzy similarity relation matrix by absolute value index method Find the classification consistent with the forecasted day Optimal ¨ and à2 LSSVM for load forecasting End Obtain a dynamic clustering graph FC YesNo Figure6.Theflowchartof theproposedforecastingmodel. 327