Model-based predictive control method for structural load reduction in wind turbines

A model-based predictive control method (MPC) for the reduction of structural load in wind turbines, comprising the steps of:

• i.—exclusively proposing a single linear internal model for the MPC, leaving aerodynamic behavior of a rotor out of such internal model throughout an operating range of a wind turbine and defining, at least:
• two inputs, a first input being aerodynamic torque that affects a rotor of the wind turbine, defined as a disturbance and, a second input being a “set-point” for generator torque, defined as a manipulated variable;
• and
• an output that allows monitoring a structural load to be reduced, defined as a control objective;
• ii.—obtaining parameters of the linear internal model from experimental data, previously measured in the turbine;
• iii.—choosing time values for control and prediction horizons, so as to balance a feedback control action and an open-loop action and to keep computational load as low as possible, wherein:
• the feedback control action is based on measurements obtained directly from the turbine during operation of an MPC controller;
• and
• the open loop action is based on a prediction of the future behavior of the system provided by the linear internal model;
• iv.—adjusting the MPC controller by:
• adjusting weights of a cost function of the MPC controller to set a degree of aggressiveness of a control action by the manipulated variable, as well as expected performance in achieving the control objective;
• and
• adjusting a state observer used by the MPC controller to set a speed of estimation of unmeasured internal states of the system, as well as a sensitivity of the observer to measurement noise of a measured angular speed of a generator and
• v. carrying out an implementation test to verify feasibility of executing a designed MPC controller on a control platform.