Application Of Ann For Short Term Load Fore Casting
This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA
Overview
Download & View Application Of Ann For Short Term Load Fore Casting as PDF for free.
More details
- Words: 451
- Pages: 1
Application of Artificial Neural Network for Short Term Load Forecasting 1
N. Amral1, D. King2, C.S. Rzveren3
PT. PLN (Persero), Indonesia, [email protected] University of Abertay Dundee, UK, [email protected] 3 University of Abertay Dundee, UK, [email protected] 2
Abstract- As accurate regional load forecasting is very important for improvement of the management performance of the electric industry, various regional loads forecasting methods have been developed. In this paper we present the development of short term load forecaster using artificial neural network (ANN) models. Three approaches have been undertaken to forecast the load demand up to 24 hours ahead. These models are applied to the South Sulawesi Electricity System and the comparative summary of their performances are evaluated through simulation.
SUMMARY Load Characteristic Following a pre-analysis of the load shapes of South Sulawesi electricity systems, it was decided that the days can be divided into four groups: weekdays (Monday to Friday), Saturday, Sunday and Holiday Description of the Model Many experiments with different ANN architectures were conducted in order to identify the architecture that gives the best results. In this paper we report the three approaches of ANN model that were implemented for the forecasting next day’s load profile of the South Sulawesi Electricity System. These three ANN approaches that mentioned above are: 1. The ANN model that has 24 output nodes to forecast a sequence of 24 hourly loads at a time. 2. The ANN model that forecast the peak and valley load and the result is used to forecast the load profile, and 3. A system with 24 separate ANN in parallel, one for each hour of the days is used to forecast the load demand. .
The load data were consisted of one series of hourly measurements of the load supplied by the utility in South Sulawesi for 2005 and 2006. The weather data are also available. Result and Observations The load profiles of 2006 (holidays are excluded) are forecasted using each developed ANN and the results are tabulated as following table. TABLE I Monthly absolute error statistics load forecast for 2006 (all figures are percentages) Month
Model #1 Mean stdev
Model #2 Mean stdev
Model # 3 Mean stdev
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec yearly
4,56 3,96 4,44 4,74 4,45 2,83 3,55 3,08 4,06 3,70 3,36 3,63 3,87
3,14 2,85 2,51 3,35 3,09 2,46 2,52 2,67 2,45 2,89 2,75 3,09 2,81
1,51 1,38 1,37 1,37 1,41 1,13 1,33 1,34 1,55 1,45 1,23 1,56 1,38
2,27 1,79 1,96 1,57 1,71 0,87 1,41 0,95 1,19 1,17 1,48 2,13 1,72
0,97 0,94 0,68 0,97 0,96 0,62 0,74 1,10 0,69 0,68 0,79 2,16 1,08
0,44 0,44 0,30 0,43 0,32 0,28 0,36 0,39 0,43 0,27 0,33 1,50 0,56
N. Amral1, D. King2, C.S. Rzveren3
PT. PLN (Persero), Indonesia, [email protected] University of Abertay Dundee, UK, [email protected] 3 University of Abertay Dundee, UK, [email protected] 2
Abstract- As accurate regional load forecasting is very important for improvement of the management performance of the electric industry, various regional loads forecasting methods have been developed. In this paper we present the development of short term load forecaster using artificial neural network (ANN) models. Three approaches have been undertaken to forecast the load demand up to 24 hours ahead. These models are applied to the South Sulawesi Electricity System and the comparative summary of their performances are evaluated through simulation.
SUMMARY Load Characteristic Following a pre-analysis of the load shapes of South Sulawesi electricity systems, it was decided that the days can be divided into four groups: weekdays (Monday to Friday), Saturday, Sunday and Holiday Description of the Model Many experiments with different ANN architectures were conducted in order to identify the architecture that gives the best results. In this paper we report the three approaches of ANN model that were implemented for the forecasting next day’s load profile of the South Sulawesi Electricity System. These three ANN approaches that mentioned above are: 1. The ANN model that has 24 output nodes to forecast a sequence of 24 hourly loads at a time. 2. The ANN model that forecast the peak and valley load and the result is used to forecast the load profile, and 3. A system with 24 separate ANN in parallel, one for each hour of the days is used to forecast the load demand. .
The load data were consisted of one series of hourly measurements of the load supplied by the utility in South Sulawesi for 2005 and 2006. The weather data are also available. Result and Observations The load profiles of 2006 (holidays are excluded) are forecasted using each developed ANN and the results are tabulated as following table. TABLE I Monthly absolute error statistics load forecast for 2006 (all figures are percentages) Month
Model #1 Mean stdev
Model #2 Mean stdev
Model # 3 Mean stdev
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec yearly
4,56 3,96 4,44 4,74 4,45 2,83 3,55 3,08 4,06 3,70 3,36 3,63 3,87
3,14 2,85 2,51 3,35 3,09 2,46 2,52 2,67 2,45 2,89 2,75 3,09 2,81
1,51 1,38 1,37 1,37 1,41 1,13 1,33 1,34 1,55 1,45 1,23 1,56 1,38
2,27 1,79 1,96 1,57 1,71 0,87 1,41 0,95 1,19 1,17 1,48 2,13 1,72
0,97 0,94 0,68 0,97 0,96 0,62 0,74 1,10 0,69 0,68 0,79 2,16 1,08
0,44 0,44 0,30 0,43 0,32 0,28 0,36 0,39 0,43 0,27 0,33 1,50 0,56
Related Documents
Application Of Ann For Short Term Load Fore Casting
December 2019 6
Ann Approch For Short Term Load Foresting For Illam Region
December 2019 7
Short Term Volunteer Application
November 2019 15
Ann Based Short Term Load Forecastinf In Electricity Markets
December 2019 4
Application Form For Short Term Trips
October 2019 10