Seite - 130 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 3283 hybridmodelwehaveconstructed. Table5 is thepredictionresult composedofMLP,andMAPEis usedasameasureofpredictionaccuracyandthepredictedresultswith thebestaccuracyaremarked inbold.Asshowninthe table,overall, amodelconsistingofnineandninenodes ineachhidden layer showedthebestperformance.Althoughthenineandsixnodes ineachhiddenlayershowedabetter performance inClusterA, themodelconsistingofnineandninenodeswasselectedtogeneralize the predictivemodel. (a) Cluster A (b) Cluster B (c) Cluster C Figure5.Resultsof similar timeseriesclassificationsusingdecisiontrees. Table5.MAPEresultsof themultilayerperceptron. Cluster# NumberofNeuronsinEachLayer 9-6-6-1 9-9-6-1 9-9-9-1 ClusterA 3.856 3.767 3.936 ClusterB 4.869 5.076 4.424 ClusterC 3.366 3.390 3.205 130