Seite - 369 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1253 6.Conclusions In recent years, with the graduallyworsening energy crisis and the intensification of global warming,EVshavebecomeoneof themaindevelopmentdirections fornewenergyvehiclesdue, to their energy savings and emission reductions. EV charging stations are an important part of thepower load; thus, researchon their short-termload forecasting isnotonlyofgreat significance for economicdispatch in the grid, but also contributes to stable operationof the charging station. In thispaper, ashort-termloadforecastingmethodforEVchargingstationscombiningNILAwith CNN is established, whereNI is used to improve the optimization performance of LA, and the hybrid techniqueNILAis introducedtodetermine theoptimalparametersofCNNmodel, soas to obtainbetterpredictionaccuracy. Throughanalysisof loadcharacteristics in thechargingstation, ten influential factorsareselectedas input, includingseasonalcategory,maximumtemperature,minimum temperature,weathercondition,daytype,andthe loadsat thesamemoment inpreviousfivedays. According to the case studies, CNN integratedwithNILAoutperformsothermodels in termsof predictionprecision, indicatingthatNILA-CNNmodel isapromisingtechniqueforshort-termload forecastingofEVchargingstation. AuthorContributions:Y.L.designedthis researchandwrote thispaper;Y.H.providedprofessionalguidance; M.Z.processedthedataandrevisedthispaper. Funding: This researchwas fundedby [theFundamentalResearchFunds for theCentralUniversities] grant number [2014MS146]. Acknowledgments: Thiswork is supportedby theFundamentalResearchFunds for theCentralUniversities (ProjectNo. 2014MS146). Conflictsof Interest:Theauthorsdeclarenoconflictof interest. Abbreviations EV Electricvehicle CNN Convolutionalneuralnetwork LA Lionalgorithm NI Niche immunity NILA Lionalgorithmimprovedbyniche immunity ANN Artificialneuralnetwork SVM Supportvectormachine RBFNN Radialbasis functionneuralnetwork TS timeseries RE Relativeerror RMSE Rootmeansquareerror MAPE Meanabsolutepercentageerror AAE Averageabsoluteerror LA-CNN Convolutionalneuralnetworkoptimizedbylionalgorithm NILA-CNN Convolutionalneuralnetworkoptimizedbyniche immunity lionalgorithm References 1. Tribioli, L. Energy-BasedDesign of Powertrain for aRe-EngineeredPost-TransmissionHybrid Electric Vehicle.Energies2017,10, 918. [CrossRef] 2. Tan,K.M.;Ramachandaramurthy,V.K.;Yong,J.Y.;Padmanaban,S.;Mihet-Popa,L.;Blaabjerg,F.Minimization of LoadVariance inPowerGrids—InvestigationonOptimalVehicle-to-Grid Scheduling. Energies 2017, 10, 1880. [CrossRef] 3. Zhang,Y.; Su,X.; Yan,X.; Li,M.; Li,D.D.Amethodof charging load forecast basedonelectric vehicle time-spacecharacteristics.Electr. PowerConstr. 2015,7, 75–82. 369