Sliding mode control for doubly fed induction machine

Abstract

The objective of this thesis is to provide robust control methods for Doubly Fed Induction Generator (DFIG) mainly used for wind energy conversion systems (WECS). The controller deals with the power electronic stage needed to control the electric machine. Due to the high sensibility of this machine against disturbances the systems requires an accurate controller that allows to follow the stator power regulation despite the presence of disturbances. DFIG-based WECS are highly sensible to stator voltage perturbations because the stator is directly connected to the electric grid. The offered controller can regulate torque and reactive power even during low depth voltage dips by injecting negative sequence current without the necessity of sequence components separation of the rotor current. The proposed control system is based on Sliding-Mode Control (SMC), offering a fast-dynamic response and providing insensitivity to matched and bounded disturbance/uncertainties. The naturally discontinuous control signals of SMC can be used for direct switching of power electronic devices avoiding modulation and making the control system simple and robust. Furthermore, the controller does not depend on electric machine's parameters and it is decoupled by means of reference frame orientation. The switching frequency is analyzed using a frequency-domain method known as Tsypkin's locus, therefore it is possible to calculate a hysteresis width which will maintain the switching frequency inside safe operational range preventing to exceed the maximum switching frequency of semiconductor devices. The dynamic response of the control system under balanced and unbalanced voltage faults is enhanced by the application of demagnetizing current for fast natural flux clearance. The demagnetizing current is supplied using the original controller by modifying the torque and reactive power references. The operation of the controller and the ideas presented in this thesis have been verified through simulations made in National Instruments\textsuperscript LabVIEW and Matlab Simulink.