Seite - 323 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1449 loaddecreaseson rainyandsnowydaysonaccountof thedecelerationof e-buses,which leads to adecrease in thedailydrivingmileageandcharging timesaswell as the reductionof total load in thechargingstation. Tothisend, rainyandsnowydaysareanothervital factor thataffects the load characteristicsofe-buschargingstations. Figure4.Relationshipbetweenweatherconditionsanddaily load. 2.3. BusDispatching Theschedulingofdeparture timeandoff-runningtimeisamomentous taskforbusoperation companies. In lightof thedailyplanofbusdispatching,differentcharging intensitiesofe-buses in the stationcausechanges in thedaily loadcurve in thechargingstationatdifferentperiods.Moreover, diversedemandsof thepublic, traffic jams,andsuddensituationsrequire theadditionof temporary e-buses toenhance transport capacity,whichbringsaboutchanges inbusschedulingondifferentdays. Busdispatching isoneof thedirect reasons for thefluctuationofdaily loadcurveandthedistinction of loadcurvesamongdays. According to thedispatchplanmade inadvance, the totalnumberof e-buses thatneedtobechargedonapredicteddaycanbeestimated;namely, theaccumulatednumber ofe-buseschargeddaily,which isusedasanindicator toreflect theeffectofbusdispatchingonthe loadof thequick-changee-buschargingstation. 3.Methodology 3.1. FuzzyClustering FC analysis is a mathematical technique that achieves classification of objects through the establishmentof fuzzysimilarity relationsbasedontheir characteristics, familiarityandcomparability. Thefuzzyequivalentmatrixdynamicclusteringmethodis implemented in thispaper. Supposen samplesonthepredictedday, that isX=[x1,x2,...,xn]. Eachsamplexj comprisesm indicators, expressedasxj= [ xj1,xj2,...,xjm ]T, j=1,2,...,n. ThespecificstepsofFCcanbeexplainedas follows: (1)Datastandardization.Consideringdifferentdimensionsandordersofmagnitude, thedata mustbestandardizedasEquation(1) [27]. x′jk=(xjk−xkmin)/(xkmax−xkmin), (j=1, 2, ... ,n; k=1, 2, ...,m) (1) where xjk is the raw data, xkmin and xkmax are theminimum andmaximum of x1k,x2k, · · · ,xnk, respectively,x′jk is thestandardizeddata. (2)Establishmentof fuzzysimilarity relationmatrix. Inorder tomeasure thecomparabilityof the classifiedsamples, a fuzzysimilarity relationmatrixR= { rij } needs tobeconstructedbysimilarityof 323