Optimize control current in magnetic bearings using automatic learning control

How to solve the imbalance problem in active magnetic bearings (AMB) is always concerned. In this paper a new unbalance control method, automatic learning control, is proposed to solve this problem. Using the method, the rotor can be forced to rotate about its inertial axis. The synchronous compensation current of the AMB is optimized through iterative "learning". The learning gain and learning cycle of the controller are automatically determined according to the rotational speed. Experiments are carried out to examine its control effect at fixed rotational speed and variable speed, respectively, and results prove that the proposed compensation scheme is effective in AMB control.