Page - 384 - in Short-Term Load Forecasting by Artificial Intelligent Technologies

Energies2018,11, 1138 characteristicsofelectricityusecreateabadeffect forshort-termloadforecasting. Theeffect reduces whenweusecorrespondingtrainednetworks fordifferentcustomers. Therefore, theperformance is generally improvedbyclustering. Table7.ComparedMAPEsforninecustomers in threecategorieswithorwithoutclustering. Customer Category Feeder MAPEwithClustering MAPEwithoutClustering 37148000 1 2 10.80% 22.98% 51690000 1 7 9.25% 21.17% 37165000 1 7 11.12% 20.12% 53990001 2 2 10.07% 27.31% 54265001 2 3 11.91% 22.72% 54265002 2 3 10.76% 28.45% 31624001 3 35 13.56% 21.12% 41661001 3 34 12.23% 24.85% 76242001 3 33 9.98% 28.38% D 6DPSOLQJ 3RLQWV 6DPSOLQJ 3HULRG PLQXWHV E 6DPSOLQJ 3RLQWV 6DPSOLQJ 3HULRG PLQXWHV F 6DPSOLQJ 3RLQWV 6DPSOLQJ 3HULRG PLQXWHV G 6DPSOLQJ 3RLQWV 6DPSOLQJ 3HULRG PLQXWHV $FWXDO ORDG )RUHFDVWLQJ ORDG ZLWK FOXVWHULQJ )RUHFDVWLQJ ORDG ZLWKRXW FOXVWHULQJ Figure12.Comparedcurvesofactual loadandforecasting loadofCustomer53990001withorwithout clustering: (a–d) theresults for fourdifferentdays inNovember2013. TheinputofproposednetworkincludesDp,Wp,TpandPl,whichmeansthat thenetworkobtains and fuses theprevious load changingprocess andother environmental information. In this case, weremovedthe input layerandthe followingfullyconnected layers in thenetwork. Thecomparison results of Customer 53990001withmulti-source or only load data input are shown in Figure 13. The comparedMAPEs for nine customers in three categories fromdifferent feeders are shown in Table 8. The experimental condition is the same as the one above. It can be concluded that the performance of only using load data is obviously poorer. Although the change shape is similar to actual, the curvesdeviate from theactual curves. Correspondingly, theMAPEsare larger. The reason is thatdate,weatherandtemperaturearenecessary factors toconsiderduringshort-termload forecastingprocessing. Peoplewouldraise their loadonahotorcoldday,evenarainyorsnowyday. 384