The project was presented to leading R&D organisations in the scientific community so that the methodologies for solving the problem were varied.
It was decided to focus the efforts on improving the forecast of the current model in those cases identified as the ones with the greatest uncertainly: time change days and adjacent days, days with extreme temperatures, holidays, and Christmas period.
The project was completed in 2016
Red Eléctrica collaborated with two universities, the Miguel Hernández University in Elche and the Technical School of Industrial Engineering in Madrid.
As a result of the new combined demand prediction algorithms, a higher reliability of the prediction has been achieved, since it does not depend on a single algorithm alone, and a decrease in the mean squared error of 0.25% in the case of the electrical demand of the peninsular system, and between 0.14% and 2.28% on average in non-peninsular systems depending on the specific isolated system.
Obtaining more precise demand forecasts allows the system operator to optimise the generation resources used by the electrical system, which also indirectly reduces emissions as a result of the more efficient programming of thermal generation in the system (higher forecast accuracy, less need for reserve) and maximises the use of renewable energy.