Faults tolerant control of wind turbine based on Laguerre model predictive compensator

Abstract

In this paper fault tolerant control is performed to counterbalance actuators' faults in wind turbines using model predictive control (MPC). Laguerre MPC parameterization is used in order to improve the numerical conditioning of the MPC optimization problem. Faults in the actuators for blade pitching and generator torque control are considered. For each fault, a step of fault detection and estimation is done followed by fault compensation. The pitch fault is characterized by a change in the actuator dynamics and is therefore estimated using a Kalman-like observer. Then, MPC compensator is employed to calculate the input that minimizes the difference between the defected sensor (for which the model is estimated using the observer) and the nominal situation (without fault). The formulation of Laguerre MPC for fault compensation is provided. The generator torque fault is an offset fault and is estimated based on the discrepancy between the measured value and the set-point. Correction of this fault is done by feed forward control. The implementation of the combined estimation and Laguerre MPC compensator and the feed forward control were found to offer a good online estimation and compensation of the fault under realistic fault scenarios in the wind turbine benchmark.