Robust Tube-Based Reference Tracking Nonlinear Model Predictive Control for Wind Turbines

Recently, a research effort to extend nonlinear model predictive control (NMPC) methods from setpoint stabilization to reference tracking has been felt increasingly. On the other hand, applying robust NMPC approaches to a larger class of systems subject to disturbances and uncertainties often results in the control approach’s complexity and a significant increase in computational burden. Therefore, this study proposes a general reference-tracking, robust tube-based NMPC for nonlinear systems. The system is subject to input– and state–dependent uncertainties. Robust recursive feasibility and constraint fulfillment are guaranteed by the employment of suitable terminal components. The proposed control approach includes two algorithms for offline and online computations. In the offline algorithm, the tube shape is formed based on the stabilizing feedback and the Lyapunov function. In the online implementation, only the bounds related to these functions are used, which makes the proposed controller easy to implement. In addition, to facilitate efficient real-time implementation, a parallel Newtonian-type approach is applied to solve the optimization problem of NMPC. The proposed control scheme is designed to control wind turbines in the presence of additive and parametric disturbances and is compared with the nominal tracking NMPC scheme and two other newly developed methods. Simulations are performed for fatigue and extreme load scenarios, and results show that the proposed controller has superior robust performance compared to the other methods.
Note to Practitioners
—Model predictive control is a widely used industry method for advanced process control. In the field of robust MPC, tube-based MPC provides a suitable compromise between complexity and performance. In addition, in most industrial applications, the purpose of control goes beyond stabilizing set points and tends to include tracking time-varying references. Hence, this paper proposes a general reference-tracking, robust tube-based NMPC for nonlinear systems. In the offline design of the proposed controller, only the calculation of scalar bounds is needed, simplifying the online implementation of the controller. In addition, in the real-time implementation of the proposed controller, a parallel Newtonian-type method is used, which makes it more suitable for practical applications. Among all wind turbine control strategies, NMPC is emerging as a strong choice from the point of view of optimal power extraction and mechanical load reduction. Therefore, a robust NMPC approach is proposed here to track time-varying references dependent on wind speed for wind turbines exposed to external disturbances, static frictions, and model uncertainties.