Skip to Main Content
Nowadays in deregulated electricity markets, forecasting becomes more and more important. Accurate estimating price is the most essential task and the basis for any decision making and developing biding strategies. In this paper artificial neural network (ANN) is applied to forecast electricity price. ANNs have been used successfully in many forecasting applications recently. Historical data of weighted average prices in electricity market that cleared with pay as bid settlement rule and also forecasted electricity demand are used to train neural networks (NN) model. The data for historical electricity prices are considered by two criterion of temperature and load in order to improve training. Data are categorized in to two seasonal parts of warm and cold days and also they are categorized by three loads level of low, normal and peak. To support better accuracy of estimating prices six different NN architectures are modeled for combination of the criterions, the NN also selected electricity prices of similar days in previous year. In the end, the results of forecasting prices of ANNs are compared with the results obtained from regression model. There is confirmed that, AANs have better performance comparing to traditional method in forecasting electricity prices.