Seite - 87 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2019,12, 57 Therefore, thepowermanagementcanbeadjustedbytuningSOClow andSOChigh. Asa last stage, thehydrogenelectrolyzer transferredredundantenergytohydrogenstoragebasedonthe following strategies (seeFigure3b) [33]: (1) When the battery SOC was higher than 95%, the extra renewable energy was regarded asredundant. (2) Theelectrolyzermodulewouldwait for tenminutes toavoidchattering. If the total redundant energy increasedduringthisperiod, theelectrolyzerwasswitchedon. (3) When the hydrogen tankwas full or the battery SOCdropped to 85%, the electrolyzerwas switchedoff. Thus, the electrolyzerproducedhydrogenwhen thebatterySOCwasbetween85%and95%. Theelectrolyzermodulewasset toproducehydrogenatarateof1.14L/minbyconsumingaconstant powerof410W,basedontheexperimental results [33]. 2.2. InputsEnergyandOutputLoads Weappliedthehistorical irradiationandwindspeeddata [32],asshowninFigure4, to thePV andWTmodules, respectively.AsshowninFigure4, solar radiationwasabundant in thesummer but poor in thewinter; therefore, solar energy in the summer canbe stored for use in thewinter. Conversely, thewind speedwas high in thewinter but low in the summer, sowind energywas expectedtocompensate for the lackofsolarenergy in thewinter.However, thecompensationeffects werenotas significantasoriginallydesignedbecause thewindwasnot sufficientlystrongandthe energycostwasmuchhigher (seeTable2)whencomparedtootherenergysources.Note thatboth solarandwindenergywereconcentrated in thedaytime, indicating that thisenergyshouldbestored foruseatnight. ȱ (a)ȱ61Ȭdayȱradiationȱdata.ȱ (b)ȱAverageȱdailyȱradiation.ȱ ȱ (c)ȱ61Ȭdayȱwindȱspeedȱdata.ȱ (d)ȱAverageȱdailyȱwindȱspeed.ȱ /RQJ 7LPH ,UUDGLDQFH ( WLPH GD\ 6XPPHU :LQWHU 'DLO\ $YHUDJH ,UUDGLDQFH ( WLPH KU :LQWHU 6XPPHU :LQG 6SHHG /RQJ 7LPH WLPH GD\ 6XPPHU :LQWHU :LQG 6SHHG 'DLO\ $YHUDJH WLPH KU 6XPPHU :LQWHU Figure4.Radiationandwinddata. 87