Machine learning for detecting damage in wind turbines

Abstract

[EN] During the last years the humankind realized the importance of the environment, not only that, but also the impact that our presence has on it. This impact is meanly bad, so during the last two decades we started to implement solutions to avoid or even reverse these consequences. One of most important changes was, and still is, the renewable energies, such as the wind power. Wind power production uses the speed of wind to move the helixes of a fan that moves a turbine generating electricity power. However, since these machines are mechanical instruments and they are exposed to different conditions such as weather conditions including temperature, humidity… And, mechanical stress, due to the movement and the structure of the fans themselves. They will always have damaging processes that could lead to future failures on the engines or the structures. The scope of the thesis is to detect and predict the possible failures on the wind turbines. The main goals were met, the accuracy of the models is relevant and the implications on the implementation on real life are significant since it helps to predict some of the failures. But there is still space for improvement since the analysis shows possible new paths to research.

[ES] El proyecto consiste en desarrollar algoritmos de inteligencia artificial y aplicarlos al estudio de las turbinas eólicas para detectar posibles daños presentes y futuros.