Page - 129 - in Short-Term Load Forecasting by Artificial Intelligent Technologies
Image of the Page - 129 -
Text of the Page - 129 -
Energies2018,11, 3283
to reflectall timevariables. Figure5showstheresultof thesimilar timeseries recognition foreach
clusterusing thedecision tree. Here, samples indicate thenumberof tuples ineach leaf. The total
numberof samples is 1095, sinceweare considering thedaily consumptiondataover threeyears.
Valuedenotes theclassificationvalueof thesimilar timeseries. Table4showsthenumberofsimilar
timeseriessamplesaccordingto thedecisiontree for2016and2017.
Table3.MAPEresultsofLSTMnetworks.
TimeStep ClusterA ClusterB ClusterC Average
1 9.587 6.989 6.834 7.803
2 9.169 6.839 6.626 7.545
3 8.820 6.812 6.463 7.365
4 8.773 6.750 6.328 7.284
5 8.686 6.626 6.191 7.168
6 8.403 6.695 5.995 7.031
7 8.405 6.700 6.104 7.070
8 8.263 6.406 5.846 6.839
9 8.260 6.583 5.648 6.830
10 8.286 6.318 5.524 6.709
11 8.095 6.438 5.666 6.733
12 8.133 6.469 5.917 6.840
13 7.715 6.346 5.699 6.587
14 7.770 6.263 5.399 6.477
15 7.751 6.139 5.306 6.399
16 7.561 5.974 5.315 6.283
17 7.411 5.891 5.450 6.251
18 7.364 6.063 5.398 6.275
19 7.466 6.089 5.639 6.398
20 7.510 5.892 5.627 6.343
21 7.763 5.977 5.451 6.397
22 7.385 5.856 5.460 6.234
23 7.431 5.795 5.756 6.327
24 7.870 6.089 5.600 6.520
25 7.352 5.923 5.370 6.215
26 7.335 5.997 5.285 6.206
27 7.405 5.479 5.371 6.085
28 7.422 5.853 5.128 6.134
29 7.553 5.979 5.567 6.366
30 7.569 5.601 5.574 6.248
Table4.Similar timeseriespatterns.
Pattern ClusterA ClusterB ClusterC
2016 2017 2016 2017 2016 2017
1 62 62 62 62 62 62
2 14 14 14 14 140 138
3 107 111 107 111 20 20
4 64 58 64 58 25 25
5 14 15 14 15 1 2
6 53 52 53 52 10 9
7 16 16 5 5 99 98
8 36 37 47 48 9 11
Total 366 365 366 365 366 365
Thepredictiveevaluationconsistsof twosteps. Basedontheforecastmodelsof randomforest
andMLP,weused the trainingset from2013 to2015andpredicted theverificationperiodof2016.
Theobjectivesare todetectmodelswithoptimalhyper-parametersandthentoselectmodelswitha
betterpredictiveperformance insimilar timeseries.Next,weset the trainingset to includedata from
2013to2016andpredictedthe testperiodof2017.Here,weevaluate thepredictiveperformanceof the
129
Short-Term Load Forecasting by Artificial Intelligent Technologies
- Title
- Short-Term Load Forecasting by Artificial Intelligent Technologies
- Authors
- Wei-Chiang Hong
- Ming-Wei Li
- Guo-Feng Fan
- Editor
- MDPI
- Location
- Basel
- Date
- 2019
- Language
- English
- License
- CC BY 4.0
- ISBN
- 978-3-03897-583-0
- Size
- 17.0 x 24.4 cm
- Pages
- 448
- Keywords
- Scheduling Problems in Logistics, Transport, Timetabling, Sports, Healthcare, Engineering, Energy Management
- Category
- Informatik