Variable selection for wind power prediction using particle swarm optimization

Wind energy has an increasing influence on the energy supply in many countries, but in contrast to conventional power plants it is a fluctuating energy source. For its integration in the electricity supply structure it is necessary to predict the wind power hours or days ahead. There are models based on physical, statistical or artificial intelligence approaches for the prediction of wind power. In this paper a new short-term prediction method is described based on variable selection using particle swarm optimization and nearest neighbour search. As input variables for this prediction method weather data of a numerical weather prediction model and measured power data from wind farms of several locations in a spread area are used. Additionally a prediction model based on neural networks is described and the results of the new method are compared to the results of the neural network approach. As a result we get a reduction of the prediction error by using the new prediction method. An additional error reduction is possible by using the mean model output of the neural network model and of the nearest neighbour search based prediction approach