Grey-Box Identification of a Cone-Shaped Active Magnetic Bearing System

An Active Magnetic Bearing (AMB) is a mechatronic system that supports a rotating shaft by means of magnetic levitation. Cone-shaped Active Magnetic Bearings (AMB) can control the rotor motion in the axial and radial directions simultaneously with two bearings in a cone-shaped magnetic core. Although this configuration eliminates a dedicated axial actuator, it complicates the control due to the coupling of the axial and radial actions. Cone-shaped AMBs are nonlinear and unstable in open-loop. However, obtaining a linear plant model is usually preferred to facilitate the control design, as it can exploit classical control theory tools. Nevertheless, the control performance can deteriorate when the plant model is affected by uncertainty. In this context, this paper presents the identification of a linear and multi-variable cone-shaped AMB system using grey-box modeling. Firstly, the plant linear and nonlinear models are presented. Since the cone-shaped AMB has an unstable open-loop nature, a Linear Quadratic Regulator (LQR) is used to stabilize the plant. An experimental campaign is performed to obtain the plant response while applying persistent disturbance currents on each electromagnet. These measurements are then used to carry out the grey-box identification procedure. It is demonstrated that the identified model yields an experimental match improvement up to 10% when compared to the nominal model. This estimated model can be exploited in model-based control approaches to improve system performance