Predictive control of doubly fed induction generator used for wind energy

Abstract

The parameters variations of models systems oblige us to find and develop a robust control in contrast the classic controller (PID) because It does not satisfy the models performances, for that we must choose a mode of control that has the best response even if the system is non linear or parameters are changing In this thesis we recall generalized predictive control (GPC) and adaptive control with reference model (MRAC) because of the large industrial use of those modes of controls for a doubly fed induction generation (DFIG) connected directly to the grid by its stator and controlled by its rotor by two level converter used for wind turbine. Then we use neural network (RN) to develop the classical controller to have the same performance of predictive and adaptive control with reference model. Results by Matlab/Simulink shows very good responses in terms of reference tracking, robustness and Current harmonics distortion.