Identification of Variables of a Floating Wind Turbine Prototype

Abstract

In this paper, using real data from a low scale prototype of a wind turbine, different models have been obtained based on machine learning techniques. These models have been shown to be useful to forecast some key statistical metrics of the dynamics of the wind turbine. The models are dependent on the wind speed and the blade pitch angle. These models can be used to develop a digital twin of the wind turbine and predict its behavior, even for wind speed and pitch angles outside the ranges used for training the system.