Open AccessAn Evaluation on Diverse Machine Learning Algorithms for Hourly Univariate Wind Power Prediction in the Arctic
Author(s) -
Hao Chen,
Yngve Birkelund
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2141/1/012016
Subject(s) - wind power , univariate , arctic , meteorology , greenhouse gas , environmental science , computer science , wind power forecasting , wind speed , algorithm , power (physics) , machine learning , electric power system , engineering , multivariate statistics , geography , ecology , physics , quantum mechanics , electrical engineering , biology
Wind power forecasting is crucial for wind power systems, grid load balance, maintenance, and grid operation optimization. The utilization of wind energy in the Arctic regions helps reduce greenhouse gas emissions in this environmentally vulnerable area. In the present study, eight various models, seven of which are representative machine learning algorithms, are used to make 1, 2, and 3 step hourly wind power predictions for five wind parks inside the Norwegian Arctic regions, and their performance is compared. Consequently, we recommend the persistence model, multilayer perceptron, and support vector regression for univariate time-series wind power forecasting within the time horizon of 3 hours.