Triple seasonal methods for short-term electricity demand forecasting

Abstract

Online short-term load forecasting is needed for the real-time scheduling of electricity generation. Univariate methods have been developed that model the intraweek and intraday seasonal cycles in intraday load data. Three such methods, shown to be competitive in recent empirical studies, are double seasonal ARMA, an adaptation of Holt-Winters exponential smoothing for double seasonality, and another, recently proposed, exponential smoothing method. In multiple years of load data, in addition to intraday and intraweek cycles, an intrayear seasonal cycle is also apparent. We extend the three double seasonal methods in order to accommodate the intrayear seasonal cycle. Using six years of British and French data, we show that for prediction up to a day-ahead the triple seasonal methods outperform the double seasonal methods, and also a univariate neural network approach. Further improvement in accuracy is produced by using a combination of the forecasts from two of the triple seasonal methods.

Combining day-ahead forecasts for British electricity prices

“…Many research papers have suggested that combining will perform better than individual methods (Clemen, 1989;Clements and Hendry, 1998;de Menezes et al, 2000;Riedel and Gabrys, 2005;Altavilla and De Grauwe, 2006;Timmermann, 2006;Chen and Yang, 2007;Clark and McCracken, 2009), including some applications to electricity demand forecasting (see Taylor and Majithia, 2000;Taylor, 2010). In the context of electricity prices, García-Martos et al (2007) similarly advocate combining, but within a single model class (ARIMA), to deal with specification uncertainty.…”

Cluster-based aggregate forecasting for residential electricity demand using smart meter data

“…Many techniques for energy consumption prediction have been inspired by research on statistical and machine learning, from Linear Regression [16], [30], ARMA [18], [34], and Generalized Additive Models [4], [10], [41] to Neural Networks [3], [15], [23] and Support Vector Regression [9], [32]. However, these techniques have been typically used at very large space scales, such as predicting the electrical load of a market segment serving thousands of customers or even an entire country.…”

A time-series compression technique and its application to the smart grid

“…In recent research, an impressive number of forecasting techniques has been developed particularly for energy demand [1,7,33]. The author in [40] shows that the so-called exponential smoothing technique behaves particularly well in the case of short-term energy demand forecasting. The main idea behind the exponential smoothing techniques is the representation of any point of the time series as a linear combination of the past points with exponentially decaying weights [25].…”