NON-PARAMETRIC WIND POWER INTERVAL PREDICTION METHOD BASED ON CHANCE CONSTRAINED EXTREME LEARNING MACHINE

摘要

The present invention relates to the field of renewable energy power generation prediction. Disclosed is a non-parametric wind power interval prediction method based on chance constrained extreme learning machines. The method combines extreme learning machines with chance constrained programming models, ensures, by means of chance constraints, that the interval coverage is not lower than the confidence level, and by using the minimization of interval width as a training target, prevents dependence on probability distribution assumptions or constraints on the quantile of interval boundary, thereby directly constructing a prediction interval having high reliability and sharpness. The present invention also provides a differential convex optimization-based binary search algorithm, thereby enabling the efficient training of chance constrained extreme learning machines.