Open AccessApplication of fuzzy multiplexing of learning Gaussian processes for the interval forecasting of wind speed
Author(s) -
Alamaniotis Miltiadis,
Karagiannis Georgios
Publication year - 2020
Publication title -
iet renewable power generation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.005
H-Index - 76
ISSN - 1752-1424
DOI - 10.1049/iet-rpg.2019.0538
Subject(s) - interval (graph theory) , wind speed , computer science , wind power , gaussian process , autoregressive model , variance (accounting) , probabilistic forecasting , gaussian , wind power forecasting , mathematical optimization , statistics , meteorology , mathematics , artificial intelligence , electric power system , power (physics) , engineering , geography , physics , accounting , combinatorics , quantum mechanics , electrical engineering , business , probabilistic logic
Robust forecasting of wind speed values is a key element to effectively accommodate renewable generation from wind in smart power systems. However, the stochastic nature of wind and the uncertainties associated with it impose high challenge in its forecasting. A new method for forecasting wind speed in renewable energy generation is introduced in this study. The goal of the method is to provide a forecast in the form of an interval, which is determined by a mean value and the variance around the mean. In particular, the forecasting interval is produced according to a two‐step process: in the first step, a set of individual kernel modelled Gaussian processes (GP) are utilised to provide a respective set of interval forecasts, i.e. mean and variance values, over the future values of the wind. In the second step, the individual forecasts are evaluated using a fuzzy driven multiplexer, which selects one of them. The final output of the methodology is a single interval that has been identified as the best among the GP models. The presented methodology is tested on the set of real‐world data and benchmarked against the individual GPs as well as the autoregressive moving average model.