Forecasting the electrical energy price in Iran power market: A comparison between single and multi hour models

This paper discusses the electrical energy price forecasting in Iran power market. Due to the day-time variations in load and thereby electrical energy price, it is wise to use different models for forecasting energy price at different hours. In this paper, three different single-hour models are used to forecast electricity price at off peak, plateau, and peak load. A 24-hour model is also used to forecast electricity price of all hours simultaneously. Analysis of autocorrelation and partial autocorrelation functions suggests different models for each single hour model as well as the 24-hour model. The best models for off peak, plateau, and peak load are obtained to be ARIMA(1,1,1), ARIMA(2,1,1) and ARIMA(0,1,1), respectively. In addition, the time-series analyses result in an AR(2) model with 24-hour period for the 24-hour model as the most suitable model. The models are compared from viewpoints of accuracy and time consuming. The comparison shows that the user should compromise between accuracy and speed, when selecting single-hour or 24-hour models.