Forecasting Uncertainty Related to Ramps of Wind Power Production

Abstract

The continuous improvement of the accuracy of wind power forecasts is motivated by the increasing wind power integration. Today forecasters are challenged in providing forecasts able to handle extreme situations. This paper presents two methods focusing on forecasting large and sharp variations in power output of a wind farm called ramps. The fi rst one provides probabilistic forecasts using large temporal scales information about ramps. The second method uses ensembles to generate con dence intervals allowing to better estimate the timing of ramps. The two methods are tested and results are given for a real case study.