Active vibration damping in a magnetic levitation system using LMS algorithm

In this paper control of an unstable magnetic levitation system is studied. The system consists of a shaft suspended on two magnetic bearings, each controlled in two axes. Firstly, the tests are carried out on a discrete SISO model of the MBC500. The model of MBC500 was acquired in identification experiments described in authors former articles. Although MBC500 is a MIMO system it is possible to describe the plant as a multiple SISO model. The high order discrete model includes vital dynamics details like mechanical resonances. As other control experiments have proven, the MBC500 can be controlled with independent SISO controllers. As MBC500 is a model of the turbine, it is mostly disturbed with sinusoidal signals. The disturbance frequency is strongly correlated with the angular velocity of shaft rotation. The aim of this paper is to verify weather the system can be successfully controlled using an adaptive approach. The FXLMS algorithm implemented in the Internal Model Control structure is confronted with Weighted Minimum Variance control. Extensive simulation and laboratory tests of the LMS-IMC algorithm are carried out on the MBC500 system. The unit is spun to different angular velocities and algorithm performance is evaluated.