Identification and Control of Flexible Structures Supported on Active Magnetic Bearings

Abstract

The focus of this thesis is the development and demonstration of adaptive control algorithms for Active Magnetic Bearings AMBs) that account for uncertainties associated with structural vibration modes. A flexible rotor AMB test rig is designed and modeled using Finite Element Analysis (FEA). The flexible modes and equations of motion are obtained using modal analysis techniques. Typically, such analyses accurately predict the low frequency modes, but accuracy at higher frequencies is greatly reduced. Adaptive control algorithms that account for parametric uncertainties and unmodeled dynamics are developed to levitate and control the flexible rotor. Model parameters for the system are identified using recursive least squares, and used to update a self-tuning pole-placement controller. This self-tuning approach dramatically improves the performance and robustness of the magnetic levitation system, as demonstrated through simulations and experimental demonstrations. The flexible rotor AMB test rig is fabricated to experimentally validate the identification and control algorithms. Results clearly demonstrate the performance and limitations of the adaptive control design.