Observer-based Self-Sensing Techniques for Hybrid Active-Passive Self-Bearing Machines

Passive magnetic levitation offers advantages in terms of compactness, reliability and cost thanks to the absence of position sensors, controllers and power electronics dedicated to the rotor suspension. Among these passively levitated systems, those based on a self-bearing machine relying on electrodynamic effects show an operating speed range limited to the high speeds given that the restoring forces are created by induced currents. To address this issue, hybrid active-passive actuation approaches have recently been introduced and consists in actively controlling the axial position of the rotor through the direct-axis component of the currents flowing in the combined winding until reaching the threshold speed beyond which passive levitation can be achieved. However, this active operation requires the addition of an axial position sensor, affecting the benefits related to passive suspension. In this context, this paper proposes self-sensing techniques relying on state observers to estimate the rotor axial position and speed on the basis of the machine electromechanical model. Their performance and robustness are then assessed by means of dynamic simulations.