Short term wind power prognosis with different success criteria

The development and exploitation of wind energy is crucially dependent on short term prognoses for power from wind. Intense efforts have been applied to development and implementation of such prognoses. Traditionally the criterion for a good prognosis is that the prognosis error in terms of the sum of squares of the errors is small. The key point in the paper is the demonstration that a variety of criteria exists. Which one is relevant can not be determined generally but will depend on the circumstances of the organization owning the wind mills and applying the prognosis, as well as the general condition of the electricity system in question. For instance a company responsible for the power balance in a system (could be a system operator) might be interested in identifying the most probable wind power production, while a production company would be interested in the prognosis that will maximize earnings from wind power production. The two prognoses need not coincide. The consequences of this are illustrated by application to the integration of wind power in the Nordpool area. Using a regression analysis the prices of regulating power will be estimated. Then power curves may be estimated using wind speed production from a numerical weather prediction model from the Danish Meteorological Institute and the corresponding short term prognoses of wind power will be elaborated. From wind power production measurements the errors may be calculated. Combining this information it is possible to find the consequences of using different prognosis criteria and hence ultimately to identify the one(s) that is (are) relevant in a specific context