Several schemes have recently been proposed for achieving either fault tolerance or self-sensing in magnetic bearings. The present work describes the fundamental connection between ability to actuate and ability to sense in a partially failed magnetic bearing system. This relationship is then exploited to construct a self-sensing scheme which operates in the presence of detectable actuator or amplifier faults. Such an approach is advantageous in fault tolerant systems because it reduces or eliminates the need to address potential independent failure mechanisms in sensors and actuators.
Based on a model reference parameter estimation mechanism, the self-sensing scheme is shown to provide acceptable position measurement accuracy and bandwidth under various actuator/amplifier faults which are actuator tolerable. Estimates of increase in noise floor and loss of bandwidth under fault conditions are provided. The issue of estimator convergence under fault conditions is examined in detail with comments on implementation complexity arising from scheduling convergence control on fault state.