Contribution à la modélisation de paliers magnétiques actifs auto-détecteurs

Active magnetic bearings with a self-sensing approach are electromechanical transducers that realize two different functions at the same time: guidage function , and sensor function for the position of rotor. In this thesis, we developed an electromagnetic model for conception and optimization of these active magnetic bearing. In order to achieve this goal, different models were developed during this thesis: - An equivalent high frequency permeability of a laminated and polarized material builds with a reversible permeability measure and an homogenization model for high frequency; - A simulation model induced currents in a laminated rotor created by its own rotation. This one permits eddy current losses determination and the impact of rotation on the position estimation; - Finally, a high frequency impedance model for a laminated, saturated and polarized magnetic circuit. This model computes in 3 steps: nonlinear magnetostatic calculation, integration of high frequency model and linear magnetodynamic calculation. These different approaches were validated experimentally with an Epstein frame and with two different magnetic bearings. All these models were then exploited in a tool for optimal conception of active magnetic bearings with a self-sensing approach. Finally, a bi-objective multivariable optimization was realized to design an optimal active magnetic bearing to both estimate position and create force.