The first method involves forecasting winds at each turbine at a given farm. The wind to power conversion is performed on a per-turbine basis and the resulting turbine powers are summed to produce an overall power forecast at the given farm. The second method involves utilizing a mean wind forecast for the entire farm. The wind to power conversion is performed by modeling farm power against the mean observed winds at a given farm. Finally, the forecasted mean winds are converted to farm power using the mean wind to farm power model.
This paper will compare these two methods and will include information on the wind forecast being used, as well as the methodology of how the wind to power conversions are created using a data mining technique that utilizes turbine level wind observations, power observations and total farm power observations. The paper will conclude with a comparison of the forecasting error of these two power prediction methods.