Periodically correlated models for short-term electricity load forecasting.

During the last two decades, the model developed by Cancelo and Espasa (1991) has been used for predicting the Spanish electricity demand with good results. This paper proposes a new approach for estimating multiequation models that extends the previous work in different and important ways. Primarily, 24-h equations are assembled to form a periodic autoregressive-moving-average model, which significantly improves the short-term predictions. To reduce the computational problem, the full model is estimated in two steps, and a meticulous model of the nonlinear temperature effect is included using regression spline techniques. The method is currently being used by the Spanish Transmission System Operator (Red Eléctrica de España , REE) to make hourly forecasts of electricity demand from one to ten days ahead. [ABSTRACT FROM AUTHOR]