An innovative algorithm for the power loads forecasting in Italian transmission grid: development and main results of the PREVEL software of Osmose project

Abstract

In the framework of the European OSMOSE project, the Zonal Energy Management System requires, every 15 minutes, the forecast of the loads at each node of the electricity transmission and sub-transmission network for the following three hours. The very short-term forecasts are provided using an Analog Ensemble scheme and an autoregressive method whose inputs consist of the last two months of short-term forecasts and load measurements. The required short-term forecasts result from a Random Forest algorithm trained using the last four months of meteorological data and load measurements. This article describes the first results of the ongoing trial, applied in Southern Italy.