Analysing wind turbine states and scada data for fault diagnosis

The diffusion of Supervisory Control And Data Acquisition (SCADA) systems has revolutionized the management of wind farms. Advanced performance optimization can lead to considerable improvement of power extraction, and therefore might also extend the possibilities for exploiting the wind resource. The price to pay is the challenge in elaborating vast amount of information into knowledge and visualizing them intuitively. Further challenge lies in the stochastic nature of the wind resource and in the complex mechanical structure of wind turbines: the optimization task therefore lies at the crossroad of physics, statistics, mechanical engineering, data visualization. This has led to fruitful collaboration between academy and industry, as the present work is. In this study, a mathematical and graphical method for elaborating wind turbine dynamics is formulated. Its key points are intuitiveness and versatility: the method can be used for a bird's eye view on a portfolio of wind turbines, for diagnosing and preventing fault onsets. The output doesn't depend on the nature of the single SCADA supplier and is potentially universal. In this work, some examples of applications to a wind farm owned by Renvico s.r.l are discussed.