Real-time adaptive control of active magnetic bearings using linear parameter varying models

Active magnetic bearings (AMBs) are electromechanical devices that provide noncontacting support of rotors, eliminating friction and other problems associated with typical mechanical bearings. However, these systems are highly nonlinear, inherently unstable, and have dynamic complications associated with flexible vibration modes and gyroscopic effects. Designing stabilizing, robust, real-time control algorithms for AMBs remains a research challenge that has to some extent limited their commercial viability. Typical AMB control strategies rely on linear, time-invariant dynamic models that do not account for modeling uncertainties or parameter variations. While these models are convenient, nominally accurate, and tractable due to the abundance of linear control techniques, they neglect the nonlinearities and uncertainties that make controller implementation so challenging. In this paper the authors demonstrate a real-time adaptive control strategy based on a linear parameter varying (LPV) dynamic model. System identification is used to update a state feedback controller to improve performance compared to a nominal controller. This self-tuning regulator (STR) is implemented and experimentally validated on a simple AMB test rig