Iterative Control-Relevant Identification and Controller Enhancement of a MIMO Magnetic Bearing System

he magnetic bearing systems are intrinsically unstable, and need the feedback control of electromagnetic forces with measured displacements. So the controller design plays an important role in constructing high performance magnetic bearing system. In case of magnetic bearing systems, the order of identified model can be high because of unknown dynamics included in closed loop systems - such as sensor dynamics, actuator dynamics - and non-linearity of magnetic bearings. "Identification for control"-joint optimization of system identification and controller design - is proposed to get the limited-order model which is suited for the design of high-performance controller. We applied the joint identification/controller design scheme to MIMO rigid rotor system supported by magnetic bearings. First, we designed controller of a nonlinear simulation model of MIMO magnetic bearing system with this scheme and proved its feasibility. Then, we performed experiments on M/MO rigid rotor system supported by magnetic bearings, and the results shows that the performance of the closed-loop system is gradually improved during the iteration.