|
M.Sc. student: Julian Ehlers.
This project is carried out at Risø National Laboratory in cooperation with Université du Québec à Rimouski, Canada.
This project aims to improve the understanding of the measurements required to perform effective individual blade pitch control, with the goal of enabling mitigation of structural vibrations during conditions of non-uniform inflow. A number of theoretical tools have been developed to quantify the dependence of estimator performance on the measurement signals used. We are now developing experience with these tools by designing and testing estimators on simulation models. To gain further insight into the relative importance of estimator errors on controller performance, we will test a range of estimator-based controllers, first in simulation, then ultimately in practice.
Theoretical Background
The state space models used in this work are linear and periodically-varying (LPV), in order to capture the periodicity in the equations for certain outputs. We use a Disturbance-Accommodating Control (DAC) formulation to represent a number of wind speed perturbations, including 2P spatial wind speed variations and independent wind speed steps on the individual blades. We have used distance-based control theoretic concepts to quantify the importance of modelling errors on the predictions made using simplified models of the aeroelastic turbine dynamics. Notably, we have made extensions to the linear, time-invariant (LTI) notion of the distance to unobservability and applied it to LPV systems, and we have developed metrics based on uniform complete observability to determine the modelling-error sensitivity of estimators with prescribed convergence rates.
Implementation and Simulation
The theoretical tools are implemented in Matlab, on state-space models generated by FAST. Estimators and controllers will be tested in FAST and HAWC to verify the test predictions with respect to modelling errors. A range of sensors is considered, from measurements in the tower and nacelle, to drivetrain measurements, measurements on the hub and blade roots, and aerodynamic measurements at outboard blade positions.
Supervisors: Amadou Doudou Diop(Québec), Henrik Bindner (Risø) |