Power Load Forecast System for Turkish Electric Market

Abstract

Forecasting the electric load demand in advance is very important in deregulated market conditions to give proper production, purchase, maintenance and investment decisions. Correct price forecasts also depend on accurate load prediction. In this study, effects of calendar, historical price and load data on short-term load forecast for Turkish Deregulated Electricity market are tested using feed forward neural networks. The impact of each data on forecast performance is evaluated and best performance is obtained using a combination of historical and predicted load, calendar information and historical price information