Seite - 343 - in Short-Term Load Forecasting by Artificial Intelligent Technologies
Bild der Seite - 343 -
Text der Seite - 343 -
Energies2018,11, 1282
Table1.KMOandBarlett testof sphericity.
KMO Value 0.740
Barlett testof sphericity Approximatechi-squarevalue 1525.304
Degreesof freedom(Df.) 231
Significance (Sig.) 0.000
Table2.Resultsof factoranalysis.
Indicator Variable Load ContributionRate (%)
Factor1 Minimumtemperature 0.732
35.128
Dailymaximumload 0.714
Dailyminimumload 0.726
Averagedaily load 0.870
Seasonpatterns 0.736
Peakaverage loadofpreviousday 0.922
Valleyaverage loadofpreviousday 0.801
Average loadof thedaybefore 0.917
Average loadof2daysbefore 0.830
Average loadof3daysbefore 0.695
Factor2 Maximumtemperature −0.732
19.646
Average temperature −0.697
Humidity 0.810
Visibility −0.724
Weatherpatterns 0.724
Average loadof4daysbefore 0.547
Factor3 Typeofdate 0.622
10.514Average
loadof5daysbefore 0.612
Average loadof6daysbefore 0.609
Factor4 Airpressure 0.563 7.746
Factor5 Date 0.883 6.087
Factor6 Windspeed −0.533 5.313
3.3. TheAnalysis ofCorrelation
Additionally, this paper conducted a further analysis of the correlation between the amount
ofhistorical loadand the target load fromtwodifferentviewpoints soas to eliminate the internal
correlation.Ontheonehand, thepartialautocorrelationfunction(PACF)wascarriedout throughout
the overall power load to dig out the correlation between the target load and the previous load.
Ontheotherhand, thewhole loaddatawith thesametimeintervalwerealso implementedbyPACF
individually toseektherelationshipamongthe loadwiththesametime. Theresultsofpartialauto
correlationcanbeseen inFigures4and5, respectively.
For instance,under theconfidence levelof90%, it canbeseenfromFigure4 that the lagsof the
first2haresignificant to thecurrentdata. That is tosay, the loadsof thefirst twohoursare influential
to thecurrent load.As forFigure5, it isknownthatonly thefirst lag1 isprominent to thecurrent load
dataexcept the loadof00:00 (Lag2). Consequently, it canbeconcludedthat the four factors including
thefirst twohoursbefore00:00andthesametimepower loadthatoccurredyesterdayandtheday
beforeyesterdaywereselectedas the input factorsat the timeof00:00.
343
Short-Term Load Forecasting by Artificial Intelligent Technologies
- Titel
- Short-Term Load Forecasting by Artificial Intelligent Technologies
- Autoren
- Wei-Chiang Hong
- Ming-Wei Li
- Guo-Feng Fan
- Herausgeber
- MDPI
- Ort
- Basel
- Datum
- 2019
- Sprache
- englisch
- Lizenz
- CC BY 4.0
- ISBN
- 978-3-03897-583-0
- Abmessungen
- 17.0 x 24.4 cm
- Seiten
- 448
- Schlagwörter
- Scheduling Problems in Logistics, Transport, Timetabling, Sports, Healthcare, Engineering, Energy Management
- Kategorie
- Informatik