Self-sensing Magnetic Bearings (Part II) Effects of Saturation

In our Part I paper, a nonlinear parameter estimation technique was presented by which the position of rotor supported by magnetic bearings can be deduced from the bearing current waveform. The estimator embedded an idealized model of the bearing inductance parameterized by the air gap. The performance of the estimator is limited by the quality of this embedded model. Magnetic saturation incurs a change of core reluctance which leads to an estimation error. This paper investigates how saturation affects the performance of the position estimator based upon an idealized model. Identifying the effects of saturation is a prerequisite to exploring the possibility of using self-sensing magnetic bearings in high performance applications such as jet engines. Thzs work demonstrates that, in extreme cases, there is a sign reversal of the forward gain of the estimator due to saturation. A simple model of saturation is presented. This model enables one to qualitatively evaluate the effects of saturation on the performance of the estimato!". A good agreement is observed between simulation and e1;periment. The structure of a gap sensing estimator which accounts for this saturation effect is proposed and some stability issues which arise in its implementation are explored. This solution will require identification of the saturation nonlinearity in the actuator, but promises to recover' the performance of previously reported estimators even in the face of fairly severe saturation.