Improving short term load forecasting using double seasol arima model

Abstract

Forecasting load demand with high accuracy is required to avoid energy wasting and prevent system failure. The aim of this paper is to develop a forecasting model based on double SARIMA for improving the accuracy of short term load prediction in Malaysia and compare the results with single SARIMA model. A half hourly load demand of Malaysia for 4 months, from September 01, 2005 to December 31, 2005 is used in this study with the mean absolute percentage error (MAPE) as one of the accuracy measures. The results of the identification step show that the load data have two seasonal periods, i.e. daily and weekly seasonality with length 48 and 336 respectively. The estimation and diagnostic check steps show that the best order of double SARIMA for half hourly load demand of Malaysia is ARIMA([2,3,4,8,11,16,18,19,20,21,28,29,30,32,40,41,45,46,47],1,1)(0,1,1)48(0,1,1)336 with in-sample and out-sample MAPE values of 0.96840 and 4.49251 respectively. The in-sample and out-sample MAPE of a single SARIMA model are 1.07872 and 10.45530 respectively. Thus, the current study shows that the double SARIMA model performs better than single SARIMA model since the MAPE of in-sample and out-sample are reduced by 10.22676% and 57.03126% respectively.