Prediction intervals are extremely difficult to calculate, so the team split their wind power series into an algorithm with three components to simplify the problem. Using the extreme learning machine technique, they perform the prediction on the separated components and combine them to find the optimal prediction interval.