Utilizing machine learning to predict offshore wind farm power output for European countries

No Thumbnail Available
Authors
Ozturk, Oktay
Hangun, Batuhan
Shoaeinaeini, Maryam
Issue Date
2022-10-25
Type
Conference paper
Language
en_US
Keywords
Europe , Wind power prediction , Renewable energy , Regression , Machine learning
Research Projects
Organizational Units
Journal Issue
Alternative Title
Abstract

One might assume that the types of energy resources used by a country and its level of development are related since developed nations focus on using alternative energy sources like the wind to produce green and sustainable energy for a better future. Worldwide interest in wind energy has increased over the past thirteen years. However, the expansion of wind power has made managing and planning electric power systems more challenging and complex due to its indirectness and volatility properties. Therefore, to balance the electrical power, a highly accurate wind power prediction is required. For this purpose, in this study, the prediction performance of k-Nearest Neighbour regression, random forest regression, decision tree regression, and linear regression methods are compared in detail. The data set we used in this study consists of a total of 29 different wind farms located in six European countries and data were recorded daily for 40 years. However, not all those wind farms' mechanical properties are the same, so we only selected some of the wind farms whose properties are compatible. Each data point in the data set consists of seven properties that are used to make estimations. Our results show each regressor's success in estimating the power output of wind farms.

Description
Citation
O. Ozturk, B. Hangun and M. Shoaeinaeini, "Utilizing Machine Learning to Predict Offshore Wind Farm Power Output for European Countries," 2022 11th International Conference on Renewable Energy Research and Application (ICRERA), 2022, pp. 611-615, doi: 10.1109/ICRERA55966.2022.9922823.
Publisher
IEEE
License
Journal
Volume
Issue
PubMed ID
DOI
ISSN
2572-6013
EISSN