Online identification of AMB rotor system dynamics using sliding DFT

Abstract

This paper presents an online nonparametric frequency response estimation approach for the identification of rotor dynamics of a high-speed machine supported by an active magnetic bearing (AMB) system. The closed-loop identification estimation approaches (direct and indirect) are based on a sliding discrete Fourier transform (SDFT) method that is applied in conjunction with a known multisine excitation signal design. The feasibility of the proposed identification approach is verified with experimental results on an AMB system. According to the results, the SDFT-based identification approach is applicable to online identification of rotor system dynamics in a computationally efficient manner.