การพยากรณ์ความต้องการไฟฟ้าระยะสั้นจำแนกตามประเภทผู้ใช้และพื้นที่บริการของการไฟฟ้าส่วนภูมิภาค

Abstract

This independent study has employed load research data for the short-term electricity demand forecasting in the northern zone of the Provincial Electricity Authority, which is divided into three regions. The load research data was collected every 15-minute interval over the years 2008-2012. The harmonics model is applied to the data in 2008-2011 and the forecasting result is validated with the data in 2012. It was shown that there are two approaches to forecast the load over 15-minute period on the monthly peak day. The first approach considers the historical data from the same month of preceding years, while the second approach considers the historical data from the preceding months. The forecasting errors were expressed by using the mean absolute deviation, the mean squared error, the mean absolute percentage error, and the coefficient of determination. The load research data has been adjusted to improve the forecasting performance when the forecasting errors exceed pre-specified threshold. It was found that the fifth harmonic model is proper to forecast the load on the monthly peak day. The forecasting results of Residential Service (Schedule 11) has too high errors so that the forecasting models are not acceptable. But, the forecasting results of Small General Service (Schedule 20), Medium General Service (Schedule 30), and Large General Service (Schedule 40) are acceptable. The forecasting errors of Medium General Service were minimum. The forecasting results in the PEA-N1 region were much better than those of the PEA-N2 and PEA-N3 regions.