Robust Predictive Pitch Control of a Wind Turbine

The random fluctuation in wind speed during wind power generation affects the smooth operation of the wind turbine as well as the stability of their generated power. In this work, the robust model predictive control (RMPC) is designed for a linearized wind turbine using a high-fidelity aeroelastic model (DNV-GL Bladed). The nonlinear wind turbine model is linearized into a state space form, and the wind fluctuation in the actual plant model is considered as a disturbance by the linearized model; that is, the control design model has no information about the disturbance. For the controller design, the feedback control law and robust invariant set are obtained by solving the linear matrix inequality that arises from the error system between the plant and the internal model. Then, the feedback control law and standard MPC will be combined to yield the robust predictive controller, which is applied to the plant model. The RMPC is tested and compared with the standard PI and MPC controllers in high wind speed in Matlab/SIMULINK. An improvement over the two controllers is achieved without compromising the aggressiveness and robustness of the controller.