Page 140 - IJEEE-2022-Vol18-ISSUE-1

136 | Al-Nasiri, Al-bayaty & Al-Hafidh

Fig. 6: Artificial neural network training performance.

Fig. 8: Neural network testing result.

By comparing the current research with previous research
that used the same data [17], but the previous research
predicted the total load only, while the current research
obtains the five load components each separately, to know
the properties of each component and its relationship to the
weather and the ratio of capacity in the daily load's.

The error (MAPE) for the previous search model was
4.58%. While the proposed prediction model equals 4.02%.
This means that the separation of components improves the
prediction model by 0.56%, this percentage is considered
good because the new model gives us accurate information
about each component as well as its relationship to the
weather during the year.

TABLE II

LOAD FORECASTING RESULTS

Fig. 7: Regression of the neural network No. Actual Forecast MAPE
Load Load
The trained neural network in Fig. 8. was tested with a set of
data to predict the five load compounds. The base component 1 635 628.68 0.00071
(Lighting and Domestic) always appears because unaffected
by the weather, while the summer component (Cooling) 2 590 582.59 0.00089
appears as the temperature rises and disappears when the
temperature drops, while the winter component (Air Heating 3 570 538.23 0.00398
and Water Heating) appears when the temperature drops and
disappears as the temperature rises. 4 562 546.34 0.00198
The results indicate the efficiency of the artificial neural
network in predicting weather-sensitive load components. 5 645 633.24 0.0013
Table II shows the mean absolute percentage error (MAPE)
for load prediction. The mean absolute percentage error 6 645 688.79 0.00484
(MAPE) for two weeks load forecasting is equal to (4.02%).
Figure 9 shows the two weeks load forecasting results. 7 610 607.89 0.00024

8 626 617.76 0.00094

9 535 454.67 0.01072

10 530 566.01 0.00485

11 630 634.24 0.00048

12 640 649.25 0.00103

13 570 593.47 0.00294

14 560 601.51 0.00529

Total 0.0402