Direct Closed-Loop Identification Approach to Magnetic Levitation System

It is necessary to perform the identification experiment of a magnetic levitation system in closed-loop where the unstable dynamics are stabilized by a feedback controller. However, the identification performance often suffers from the severe numerical problems in the optimization procedure to obtain the unstable property of the levitation system due to the less independent informative components, the poor excitation in the measured signals limited by the low loop's sensitivity to the external signals, the correlation between input-output of the magnetic levitation and the strong noise signal. In this work, a new identification algorithm is developed for the magnetic levitation system. It introduces the output over-sampling scheme to collect the informative input-output data, and uses the subspace characteristics to complement the excitation of the observation data. Furthermore, the new algorithm expresses the updating formula in such a manner that can easily explore the unstable dynamics of the magnetic levitation model. The results of the identification experiment illustrate the identification performance has been improved in the proposed algorithm. (C) 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.