Using Radial Basis Neural Networks to Estimate Wind Power Production

This paper compares two statistical methodologies for short-term wind power forecasting in two different wind farms located on offshore area and complex terrain respectively. The methods require as input past power measurements and meteorological forecasts of wind speed and direction (Numerical Weather Predictions or NWPs) interpolated at the site of the wind farm. Both methods include NWPs estimator modeling based on fuzzy logic and wind power forecasting modeling using neural networks combination.