Long- and short-term forecasting of the Italian electric power load is investigated on a quarter-hour time scale. The detrended and log-transformed data are modelled as the sum of a seasonal term (the potential) and intervention terms accounting for holidays. Yearly, weekly and daily periodicities are effectively captured by modelling the potential as a function defined on a 3-torus. The model yields long-term whole-year predictions, that, with a suitable correction, provide also short-term whole day predictions. When tested on Italian 2013 load data, the new one-day-ahead forecaster performs definitely better than two state-of-art forecasters currently used by the Italian TSO. © 2016 IEEE.
Long- and short-term electric load forecasting on quarter-hour data: A 3-torus approach
GUERINI, ALICE;DE NICOLAO, GIUSEPPE
2016-01-01
Abstract
Long- and short-term forecasting of the Italian electric power load is investigated on a quarter-hour time scale. The detrended and log-transformed data are modelled as the sum of a seasonal term (the potential) and intervention terms accounting for holidays. Yearly, weekly and daily periodicities are effectively captured by modelling the potential as a function defined on a 3-torus. The model yields long-term whole-year predictions, that, with a suitable correction, provide also short-term whole day predictions. When tested on Italian 2013 load data, the new one-day-ahead forecaster performs definitely better than two state-of-art forecasters currently used by the Italian TSO. © 2016 IEEE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.