Wind power prediction risk indices based on numerical weather prediction ensembles

The large-scale integration of wind generation imposes several difficulties in the management of power systems. Wind power forecasting up to a few days ahead contributes to a secure and economic power system operation. Prediction models of today are mainly focused on spot or probabilistic predictions of wind power. However, in many applications, endusers require additional tools for the on-line estimation of the uncertainty of the predictions. One solution to this is prediction risk indices, computed on wind power forecast ensembles derived from numerical weather prediction ensembles. This paper investigates the usefulness of such risk indices as a complement to usual wind power forecasts for informing on the expected level of uncertainty and the risk for large forecast errors. Results show that risk indices are useful to extract information from power ensembles and can give valuable information about the expected prediction uncertainty.