Flux position estimation in cage induction machines using synchronous HF injection and Kalman filtering

This paper proposes a new approach to the tracking of the flux position saliency of an induction machine by means of high-frequency (HF) signal injection in the synchronous rotating frame. The paper shows that existing synchronous pulsating injection methods perform poorly in the presence of secondary saliencies. A modified scheme is proposed in which a rotating HF field is established in the dq rotating reference frame. The scheme enhances the performance of harmonic elimination methods to reduce the effects of secondary saliencies. A new harmonic elimination method based on the Kalman filter is proposed in the paper. Experimental results are given showing the effectiveness of the harmonic elimination and the resulting tracking of the saturation saliency.

I. Introduction

The zero- and low-speed problems of observer-based methods for the sensorless control of ac drives are well known. The main reason is that all estimation methods, directly or indirectly, rely on the rotor induced voltage, which becomes very small at the low stator frequency. Converter effects, offsets and noise components further degrade the quality of the signal. Different solutions have been suggested in the literature to overcome this effect [19]. Recent research [20] improves the estimation accuracy of the phase angle of the flux linkage vector, which is essential for independent control of torque and flux of the machine.