Enhancement of Robustness to Inductance Variation for Rotor-Position Estimation of Doubly-Fed Induction Machine by Regressor-Vector Selection

Abstract

To implement the real-world position-sensorless drives for Doubly-Fed Induction Machines (DFIM), inductance variation is an issue of concern to practitioners. In this paper, the rotor-position estimation of DFIM by the adaptive reduced-order observer is considered, and the sensitivity to inductance variation is examined. The sensitivity analysis on rotor-position estimation reveals that the effect of inductance variation is exactly the same for all operating rotor speeds, but it varies accordingly to the operating torques; with 15% of inductance variation, the error of rotor-position estimation can be up to 5°. In addition, this sensitivity analysis paves the way to the selection of regressor vector; the magnetizing current is chosen as the regressor of the adaptation scheme in order to strengthen the robust property. The validity of the analysis and the proposed concept is verified by simulation.