Comparative study of stabilising methods for adaptive speed sensorless full-order observers with stator resistance estimation

Abstract

The full-order Luenberger flux (FOLO) observer combined with adaptive techniques is the most widely used method for sensorless vector-controlled drives in induction machines. These methods suffer inherently from an unstable regenerating region and the unobservable zero stator frequency operating point. Stator resistance estimation is also required in the low speed region leading to a complicated multiple-input multiple-output (MIMO) system, and a novel simplification of the MIMO system into a single-input single-output (SISO) system based on the two-time-scale approach is proposed for a simple stability analysis procedure. Four different stabilisation strategies are suggested for this comparison, based on the main stabilisation methods found in the literature. The `augmented error signal` and the `observer gains retuning' methods, based on the Routh-Hurwitz criterion, are used for the stabilisation of the motor speed estimator. The stator resistance stabilisation has been implemented using the `augmented error signal` method. This has been evaluated using two different gain tuning methods, one based on the Routh-Hurwitz criterion and the other on the averaging analysis technique. The different stabilising strategies are evaluated theoretically according to the simplified SISO system open-loop transfer function zeroes and compared by simulation and extensive experiments. The suggested observer is flexible for any vector controlled drive and results have been validated in a 7.5 kW experimental test bench for rotor flux-oriented vector control (RFOC) and indirect self-control drives. The stability and behaviour of the sensorless scheme are successfully verified for all operating conditions above 0.4 Hz stator frequency.