Online signal processing for accurate slotting saliency extraction using two-active SVPWM integrated excitation for sensorless induction machine control

Abstract:
Sensorless motor control approaches that use voltage step excitation rely on machine anisotropy extraction. Given the six-sector switching characteristic of space vector pulse width modulation SVPWM, it is sufficient to use one or two active inverter states to form a saliency-offset vector. Using two active states leads to more prominent saliency amplitudes related to the signal noise compared to such signals using only a single active-state. This is due to the non-deterministic characteristic of noise. For extracting saliency information from the saliency-offset vector using two active states, each sector offset is required to be accurately estimated during online operation. This paper, therefore, presents an online strategy based on a SVPWM-integrated two-active excitation that estimates all sector offsets, and shows outstanding sensorless operation, validated by measurements taken at an induction motor.