This paper presents a nonlinear variable bias controller for an active magnetic bearing (AMB). The nonlinear bearing force is analyzed theoretically and the control current for various bias current settings is derived from the nonlinear bearing equation. Then the power consumption is minimized to obtain the optimum bias current expression analytically. Results show that the optimum bias current can be calculated from the demand bearing force and the instantaneous rotor displacement. Moreover, the influences of magnetic bearing parameter errors are investigated and correction methods are introduced. Results of experimental rotational tests show that the rotor dynamics are not altered under variable bias currents if the proposed correction for parameter errors is implemented. The magnetic center of misalignment is also detected and compensated for. The proposed variable bias current controller provides not only significant energy savings, but also it is simple to implement and applicable to wide range of magnetic bearing systems without deterioration of the bearing dynamics.
Learning Adaptive Differential Evolution Algorithm from Optimization Experiences by Policy Gradient