A group of researchers from the University of Alcala (UAH) in Spain has devised a model that allows them to predict the wind speed at each turbine in a wind farm.
To develop the model, the scientists used information provided by the Global Forecasting System from the US National Centers for Environmental Prediction. The data from this system covers the entire planet with a resolution of approximately 100km.
The researchers were able to make more detailed predictions by also integrating the so-called ‘fifth-generation mesoscale model’ (MM5) from the US National Center of Atmospheric Research, which had been developed to enhance the resolution to 15km x 15km.
However, according to UAH engineer Sancho Salcedo, the information was still not sufficient to predict the wind speed accurately enough and for that reason the team turned to the use of artificial neural networks.
The networks use the temperature, atmospheric-pressure and wind-speed data provided by the forecasting models, in addition to data from the turbines themselves.
Presented with that data, the neural network system is then trained, after which it can make predictions regarding wind speed between one and 48 hours in advance.
Once the wind speed of one of the turbines is predicted, the researchers can estimate how much energy it will produce.
Wind farms are obliged by law to supply these predictions to Red Eléctrica Española: the company that delivers electricity and runs the Spanish electricity system.
The system has already been tested with good results at a wind park at Albacete.
The researchers now plan to continue improving the method. So far, results show an improvement in the prediction of a little more than two per cent over previous models.
‘Although it may seem a small amount, in fact, it is very considerable, since an improvement in the prediction of energy production can mean savings of millions of euros,’ said Salcedo.