Model Predictive Control of a Wind Turbine

Student thesis: Master thesis (including HD thesis)

  • Thomas Haugaard
  • Álvaro Martín Gómez
  • Oier Ajenjo De Torres
4. term, Control and Automation, Master (Master Programme)
In recent times, the wind turbine industry is pushing towards larger and more efficient wind turbines to decrease the cost of energy production. This requires the use of more refined control strategies able to deal with structural load reduction, while optimising power production. In order to consider multiple objectives as well as system constraints, MPC was selected as control strategy. A non-linear model of the wind turbine was derived, including the dynamics of each blade such that their pitch could be controlled independently. State estimation was performed by means of an UKF using data from GH Bladed simulation tool as system measurements. It was not possible to obtain acceptable results due to the large mismatch between the model outputs and the data from Bladed, so the first ones were used as virtual measurements by adding typical sensor noise. Afterwards, the MPC was given a prediction model derived by successive linearisation, and designed to account for both partial and full load operation of the wind turbine. The results at the present time are not satisfactory due to not having considered that the states of the linearised model are relative to the operating points in every iteration. It is believed that the results will be promising as soon as this issue is solved.
Publication date1 Jun 2022
Number of pages153
External collaboratorMita-Teknik
Engineer, Turbine Control Michael Melholt Quottrup
ID: 471937754