High frequency forecasting for wind energy using statistical and machine learning post-processing methods
Irene Schicker taks about wind forecasts needed to efficiently operate wind turbines.
With the increase in wind energy production being fed into the power grid accurate high frequency predictions of the estimate d power for the next hours and days ahead are needed to schedule feed-in rates and secure power grid stability. To achieve this a combination of different kinds of information and data sets are needed. Here, statistical and machine learning methods proved to be a suitable tool. However, a thorough selection of input data is needed as well as considering extreme events (upper and lower tails) in model training and avoiding smoothed forecasts.
A brief introduction into post-processing for wind energy applications using statistics and machine learning, including useful tools/methods/data, will be given.