System identification of MIMO magnetic bearing via continuous time and frequency response data

Abstract

The identification of multi-input multi-output (MIMO) complex systems using frequency response data has received much attention from control engineers all over the world. Magnetic bearing, as an example of a highly nonlinear, unstable and multivariable system, demands extensive research studies in control and identification. This paper focuses on identifying a magnetic bearing system using plant frequency response data. Subspace system identification methods developed using Laguerre network and an instrumental variable method in the frequency domain were used to identify a MIMO continuous time transfer function for a test-stand magnetic bearing apparatus. Simulation studies based on the experimental data show that a continuous time multivariable model was successfully identified.