Real-time random reference tracking nonlinear model predictive control: a case study on wind turbines.

Area of Science:

• Control Systems Engineering
• Renewable Energy Systems
• Stochastic Processes

Background:

• Nonlinear Model Predictive Control (NMPC) is increasingly being extended from setpoint stabilization to reference tracking.
• Uncertainty in reference signals and the need for dynamic forecasting, particularly in wind turbine control, highlight the demand for robust tracking NMPC.

Purpose of the Study:

• To propose a novel random reference tracking NMPC with dynamic forecasting of stochastic references.
• To ensure convergence to a robust invariant set through an additional constraint on the tracking stage cost function.
• To enhance the efficiency and applicability of the predictive approach using a parallel Newton-type method.

Main Methods:

• Development of a random reference tracking NMPC algorithm.
• Incorporation of dynamic forecasting for stochastic references.
• Implementation using a parallel Newton-type method for computational efficiency.
• Design specifically for wind turbine control considering random wind speed references.

Main Results:

• The proposed controller demonstrates more robust performance compared to nominal NMPC.
• The controller achieves better optimal power extraction in wind turbines.
• Significant reduction in aerodynamic loads under extreme and fatigue scenarios was observed.

Conclusions:

• The developed random reference tracking NMPC effectively handles stochastic references in applications like wind turbine control.
• The approach guarantees convergence to a robust invariant set, ensuring stability.
• The method offers improved performance in power optimization and load reduction for wind energy systems.