Examination of turbulence impacts on ultra-short-term wind power and speed forecasts with machine learning

Hao Chen*, Yngve Birkelund, Fuqing Yuan

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

6 Citations (Scopus)
18 Downloads (Pure)

Abstract

Wind turbines’ economic and secure operation can be optimized through accurate ultra-short-term wind power and speed forecasts. Turbulence, considered as a local short-term physical wind phenomenon, affects wind power generation. This paper investigates the use of turbulence intensity for ultra-short-term predictions of wind power and speed with a wind farm in the Arctic, including and excluding wind turbulence, within three hours by employing several different machine learning algorithms. A rigorous and detailed statistical comparison of the predictions is conducted. The results show that the algorithms achieve reasonably accurate predictions, but turbulence intensity does not statistically contribute to wind power or speed forecasts. This observation illustrates the uncertainty of turbulence in wind power generation. Besides, differences between the types of algorithms for ultra-short-term wind forecasts are also statistically insignificant, demonstrating the unique stochasticity and complexity of wind speed and power.

Original languageEnglish
Pages (from-to)332-338
Number of pages7
JournalEnergy Reports
Volume7
DOIs
Publication statusPublished - Nov 2021
Externally publishedYes

Keywords

  • Machine learning
  • Statistical comparison
  • Turbulence
  • Wind energy
  • Wind forecast

Fingerprint

Dive into the research topics of 'Examination of turbulence impacts on ultra-short-term wind power and speed forecasts with machine learning'. Together they form a unique fingerprint.

Cite this