Automatic learning control for unbalance compensation in active magnetic bearings

This paper proposes a new control scheme, automatic learning control, to eliminate unbalance effects, which adversely affect the operation of active magnetic bearings. This control method is based on time-domain iterative learning control and gain-scheduled control. The controller can utilize the optimal control currents for the unbalance compensations. In addition, the variable learning cycle and variable learning gain are employed in the learning process to achieve better performance against rotating speed fluctuations. The control algorithm does not require large memory size and intensive computation. We tested the control system in experiments, and the experimental results prove that the control method is effective over a wide range of operation speeds.