Identification modale de systèmes variant dans le temps à partir de réponses simulées sur un exemple d'éoliennes

Wind turbines are good examples of time-varying systems as their modal properties depend on their instantaneous configuration. To catch the variations of modal parameters in time-varying systems, classical identification methods have to be adapted to the non-stationary nature of the recorded signals. In this paper, it is proposed to study the dynamic behavior of an offshore five-megawatt wind turbine. First, a numerical model of the wind turbine is created to serve as reference. Then, the time-varying behavior of the system is evaluated by simulating a large number of possible configurations. To this purpose, time responses are generated from the numerical model submitted to different environmental conditions. The wind is considered as the main non-measured external excitation force on the structure and the responses are recorded at several locations to simulate a real measurement process. Special care is brought to the accessibility of the measurement locations and to the limited number of available sensors in practice. Using these simulated measurements, output-only identification methods are used to extract varying dynamic properties of the structure. The final objective of this work is to pave the way to online condition monitoring of wind turbines.