Closed-loop system identification of wind turbines in the presence of periodic effects

The use of trailing edge flaps has developed into a promising technique to reduce loads on large wind turbines. Fatigue and extreme loads, predominantly in the blade root, are critical to the life of rotor blades and these loads can be reduced significantly by locally influencing the lift along the span of the rotor blades. To design controllers for such “smart” rotor systems, linear models are still the tool of choice. Although linear models can be obtained from thefirst principles models implemented in aeroelastic design tools, we emphasize the value of system identification techniques. Identification of linear models of wind turbine dynamics is complicated by the fact that strong periodic components are present in output measurements. These components are associated with effects such as gravity, wind shear, skew inflow conditions, tower shadow and rotational sampling of the turbulent wind field. When traditional system identification techniques are used, the estimates may be very poor due to the strong presence of these components in the measurements. In this paper, a subspace identification method is described together with a method to remove the effect of periodic disturbances on the quality of identified models, by generating periodic signals that serve as additional inputs to the identification procedure. The paper is concluded with an example that demonstrates the effectiveness of the suggested approach.