Model Predictive Control Using Preview Measurements From LIDAR

Light detection and ranging (LIDAR) systems are able to measure conditions at a distance in front of wind turbines and are therefore suited to providing preview information of wind disturbances before they impact the turbine blades. In this study, a time-varying model predictive controller is developed that uses preview measurements of wind speeds approaching the turbine. Performance of the controller is evaluated using ideal, undistorted measurements at positions that rotate with the turbine blade and measurements obtained at the same locations, but including distortion characteristic of LIDAR systems. Using these measurements, the model predictive controller is simulated in turbulent wind conditions and its performance is compared against previously designed, linear-time-invariant H1 preview controllers and industry standard controllers. Surprisingly, even though the LIDAR distortions produce signicant measurement error, controller performance is found to surpass that obtained using individual-pitch feedback-only controllers without preview. In previous studies, errors introduced articially, but of the same order of magnitude, were shown to degrade the performance of preview control so that it is worse than using feedback only. In this study, we also incorporate a simple error model to compensate the eect of LIDAR induced error, but nd that it does not improve performance.