The Multi-Point Values of Appropriate Smoothing Parameters ? Opt . of HP-filter for Mid-Term Load Forecasting based on Neural Network

摘要

The multi-point values of an appropriate smoothing parameter of HP-filter algorithm for mid-term electricity load demand (MELD) forecasting are proposed. The case study employs the data based on the organization of the Electricity Generating Authority of Thailand (EGAT). The research shows the growth at rate of weather and economic factors influencing to the electricity demand. The main focus of the article proposes the multi-point values of smoothing parameter, and also uses the appropriate values or better smoothing parameter of HP-filter for separating the electricity load demand (kWh) signal based on preprocessing stage. The method used for forecasting is an artificial neural network. Also, these approaches show the best results of in forecasting. As the result, the multi-point values of smoothing parameters of the research can be improved the accuracy of the electricity demand forecasting.