Wide-Range Low-Frequency Inductance Identification of AMB Over Current and Position Using Variable Bias Current Excitation

Inductance is a critical yet highly variable parameter in active magnetic bearings (AMB). Accurate knowledge of inductance is essential not only for current-loop design and optimization but also for applications such as fault diagnosis and condition monitoring. However, extracting accurate inductance is challenging due to its strong dependence on both operating current and rotor position. This paper introduces a method for obtaining inductance through direct experimental identification. By integrating variable bias current control under static suspension with hundred-hertz sinusoidal signal injection, we generate a comprehensive dataset covering a wide range of current and rotor position operating points. Although variable bias current control is well-established for lower loss and fault tolerance, to our knowledge, this study is the first in the AMB field to leverage it explicitly for data enrichment and inductance identification. The resulting dataset reveals a wide range of inductance values, with the maximum nearly four times the minimum. A compact neural network was trained on this experimental data to model the inductance accurately. Experimental results show the validity of the proposed method.