Application of artificial intelligence to wind forecasting: An enhanced combined approach

摘要

Along with large-scale application of wind power, power forecasting becomes increasingly important in handling wind intermittency and integrating wind power to electric grid. This paper proposes a forecasting combination approach which makes use of the forecast results of NN (Neural Networks), SVM (Support Vector Machine), and FIS (Fuzzy Inference System) models to improve the forecast accuracy. Three types of combination methods have been tested in this paper and the one based on MSE is proved to be most effective in terms of NMAE (Normalized Mean Absolute Error) and NRMSE (Normalized Root Mean Squared Error). An improved data selection scheme is also put forward to further enhance forecast accuracy.