Very-short-term probabilistic forecasting of wind power with generalized logit-normal distributions

Abstract

Very-short-term probabilistic forecasts, which are essential for an optimal management of wind generation, ought to account for the non-linear and double-bounded nature of that stochastic process. They take here the form of discrete-continuous mixtures of generalized logit-normal distributions and probability masses at the bounds. Both auto-regressive and conditional parametric auto-regressive models are considered for the dynamics of their location and scale parameters. Estimation is performed in a recursive least squares framework with exponential forgetting. The superiority of this proposal over classical assumptions about the shape of predictive densities, e.g. normal and beta, is demonstrated on the basis of 10-min-ahead point and probabilistic forecasting at the Horns Rev wind farm in Denmark. © 2012 Royal Statistical Society.