Combining Computational Fluid Dynamics and Gradient Boosting Regressor for Predicting Force Distribution on Horizontal Axis Wind Turbine

Nikhil Bagalkot,Rune Orderløkken,Arvind Keprate

doi:10.3390/vibration4010017

Nikhil Bagalkot, Rune Orderløkken + Show 1 more

Open Access

https://doi.org/10.3390/vibration4010017

Copy DOI

Journal: Vibration	Publication Date: Mar 14, 2021
Citations: 10	License type: CC BY 4.0

Affiliation: OsloMet – Oslo Metropolitan University

Abstract

The blades of the horizontal axis wind turbine (HAWT) are generally subjected to significant forces resulting from the flow field around the blade. These forces are the main contributor of the flow-induced vibrations that pose structural integrity challenges to the blade. The study focuses on the application of the gradient boosting regressor (GBR) for predicting the wind turbine response to a combination of wind speed, angle of attack, and turbulence intensity when the air flows over the rotor blade. In the first step, computational fluid dynamics (CFD) simulations were carried out on a horizontal axis wind turbine to estimate the force distribution on the blade at various wind speeds and the blade’s attack angle. After that, data obtained for two different angles of attack (4° and 8°) from CFD acts as an input dataset for the GBR algorithm, which is trained and tested to obtain the force distribution. An estimated variance score of 0.933 and 0.917 is achieved for 4° and 8°, respectively, thus showing a good agreement with the force distribution obtained from CFD. High prediction accuracy and less time consumption make GBR a suitable alternative for CFD to predict force at various wind velocities for which CFD analysis has not been performed.

Highlights

The computational fluid dynamics (CFD) results like velocity profile, pressure distribution, and force will be discu this will be followed by outcomes of the machine learning
The study focuses on demonstrating the ability of machine learning algorithm (gradiThe study focuses on demonstrating the ability of machine learning algorithm to predict the wind turbine response to a combination dient boosting regressor in this case) to predict the wind turbine response to a combinaof wind speed, angle of attack, and turbulence intensity when the air flows over the rotor tion of wind speed, angle of attack, and turbulence intensity when the air flows over the blade
The focus and novelty of the present study is in the machine learning part

Summary

Introduction

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. The Paris agreement [1] entered into force in November 2016, where a need for an effective and progressive response to climate change is recognized.

Methods

Results

Conclusion