Wind turbine predictive control focused on the alleviation of mechanical stress over the blades

This paper presents a methodology of design of a supervisory nonlinear model predictive control (NMPC) for maximizing the relationship between two opposite variables of a wind turbine (WT): generated power and mechanical stress over the blades. The predictive model used in the NMPC structure is an integrated aeroelastic model consisting of blade element momentum (BEM) and thin-walled beam (TWB) theory. The BEM/TWB model allows realtime determination of the output power and the normalized mechanical stress of the blades. The NMPC processes measurements of the operation of the WT to calculate the set-points to be sent to distributed controllers (DCs) operating in the WT. The proposed approach is compared with a baseline controller, which maximizes power extraction from the WT. Simulation results show significant improvements in the reduction of the mechanical stress while keeping the power extraction near its maximum in the entire range of operation of the WT.