SHORT-TERM ELECTRICITY LOAD FORECASTING

Loading...
Thumbnail Image
Date
2011
Journal Title
Journal ISSN
Volume Title
Publisher
Journal of Sustainability Science and Management
Abstract
The purpose of this study was to develop the best model for forecasting Malaysia load demand. In this study, a half-hourly electricity load demand of Malaysia for one year, from September 01, 2005 until August 31, 2006 measured in Megawatt (MW) was used. The doubleseasonal ARIMA model was considered due to the existence oftwo seasonal cycles in the load data. Analysis was done using SAS package. The best model was selected based on the mean absolute percentage error (MAPE), autocorrelation function (ACF) and partial autocorrelation function (PACF) plots. The ARIMA(O,1,1)(0,1,1f'(O, 1,1)336 with in-sample MAPE of 0.9906% was the best model Comparing the one-step and k-step ahead out of sample forecasts performance, the MAPE for the one-step ahead out-sample forecasts from any horizon were all less than 1% . It can be concluded that the one-step ahead out-sample forecasts were more accurate. There was a reduction in MAPE percentages for all lead time horizons considered, ranging 89% to 96%. Furthermore, a time-series plot of out-samples of actual load data, k-step ahead and one-step ahead out-sample forecasts showed that one-step ahead out-sample forecasts followed the actual load data more closely than k-step ahead out-sample forecasts. The ACF and PACF plots must be considered in proving the best model for load demand and one-step ahead out-sample forecasts in forecasting load, especially in Malaysia load data.
Description
Keywords
Load forecasting, double seasonal ARIMA model, ACF and PACF plots, one-step ahead forecasts and k-step ahead forecasts
Citation
Collections