Improving the Model for Energy Consumption Load Demand Forecasting

  • Bunnoon Pituk
    Electrical Engineering Department, Engineering Faculty, Prince of Songkla University
  • Chalermyanont Kusumal
    Electrical Engineering Department, Engineering Faculty, Prince of Songkla University
  • Limsakul Chusak
    Electrical Engineering Department, Engineering Faculty, Prince of Songkla University

この論文をさがす

抄録

This paper proposes an application of a filter method in preprocessing stage for mid-term load demand forecasting to improve electricity load forecasting and to guarantee satisfactory forecasting accuracy. Case study employs the historical electricity consumption demand data in Thailand which were recorded in the 12 years of 1997 through to 2007. The load demand forecasted value is used for unit commitment and fuel reserve planning in the power system. This method consists of a trend component and a cyclical component decomposed from the original load demand using the Hodrick-Prescott (HP) filter in the preprocessing stage and the forecasting of each component using Double Neural Networks (DNNs) in the forecasting stage. Experimental results show that with preprocessing before forecasting can predict the load demand better than that without preprocessing.

収録刊行物

参考文献 (37)*注記

もっと見る

詳細情報 詳細情報について

問題の指摘

ページトップへ