Active robust control of wind turbines

Abstract

The research work conducted in this thesis focuses on robustness of wind energy conversion system with respect to faults in pitch actuator in order to prevent unnecessary emergency shutdown, and keep the turbine operational without significant inefficiency in its overall performance. The objective is to investigate the feasibility of using a fault estimator and a light detection and ranging (LIDAR) system as additional sensors to design a suitable control system for wind turbines. Robust control technique is used to address these issues. Three controllers are proposed in this work that try to address sources of inaccuracy in wind turbine operation: An active fault tolerant controller is first designed using a fault estimator. It is shown that a set of locally robust controllers with respect to the fault, together with a suitable smooth mixing approach, manages to overcome the problem of faults in the pitch actuator. To address the wind-dependent behavior of turbines, a second controller is designed using the LIDAR sensor. In this configuration, LIDAR provides the look ahead wind information and generates a smooth scheduling signal to provide active robustness with respect to the changes in wind speed. Lastly, utilizing both the fault estimator and LIDAR, a 2-dimentional wind-dependent active fault tolerant controller is developed to control the wind turbine in region 3 of operation. The feasibility of the proposed ideas is verified in simulation. For this purpose, the US National Renewable Energy Laboratory's FAST code is used to model the 3-balded controls advanced research turbine. A discussion on practical considerations and ideas for future work are also presented.