Application data for electricity load forecasting models

Abstract

This paper used artificial neural networks (ANN) in medium-term energy forecasting for the Metropolitan Electricity Authority (MEA) area of Bangkok, Thailand. This method could improve the electricity load efficiency of the MEA. Moreover, The combined ANN with the GIS in next paper have a key role in the decision-making for investment in new substation and power system planning for maintenance and operation. The input data were clustered by K-means algorithms before training by forecasting the models. In this research, the energy forecasting models were the ANN (2 hiddens 4 hiddens). The prediction was based on the MEA's electrical energy history (six months; three months) and the gross domestic product (GDP). The results appeared to indicate that the prediction of ANN 4 hiddens (Classified Data Input) is more accurate.