Abstract
Doubly-fed induction generator (DFIG) have been widely accepted as ideal machines for wind power generation due to their rigidness in structure and flexible output ability. Sensorless control would further improve system reliability by eliminating the needs for position sensors. In this paper, a back-EMF-based model-reference adaptive system (MRAS) sensorless control scheme is proposed for grid-connected DFIGs. Both of the reference model and the adaptive model of MRAS are first derived based on the stator-side and rotor-side dynamic models of the DFIG. Stability and convergence of the proposed controller are then verified by applying Popov's criterion. The adaptation mechanism is also obtained along the derivation, and the speed estimator is therefore completed. The MATLAB/Simulink software is utilized to establish the simulation of the sensorless MRAS controlled DFIG system. Moreover, an experimental platform is constructed with the xPC real-time hardware control suit, which achieves the proposed MRAS sensorless control of a 2.2kVA DFIG. © 2013 IEEE.