Modelling of a one-degree-of-freedom self-sensing active magnetic bearing

Abstract

This paper investigates the modelling of a one degree-of-freedom self-sensing active magnetic bearing system. Active magnetic bearings support a rotor without a mechanical contact by using electromagnetic forces to levitate the rotor. Such a system removes friction thus removing the need to consider wear and lubrication. Conventional active magnetic bearing systems require a dedicated displacement sensor to measure the position of the rotor. These sensors, however, increase the size, cost and complexity of such a system. The purpose of this thesis is to model a self-sensing alternative where the rotor displacement is estimated from the electrical behaviour of the magnetic actuator instead of being measured directly.

The work first developed a nonlinear dynamic model of a rotor suspended between two opposing electromagnets. This model knows its own rotor position and adjusts itself accordingly. A second model is then developed which employs a flux-based self-sensing system and a Luenberger observer to estimate rotor displacement.

These two models are subjected to two loadcases with one case featuring an excitation force and no initial displacement and the second case featuring the opposite. The results show the models ability to remarkably follow and stabilise the system under in the self-sensing control.