Elnaz Vahedforough

Global Adaptive Compensation of Sinusoidal Disturbances in Active Megnetic Bearing Systems.

Date: Wednesday July 18th, 2007

Abstract:

Active Magnetic Bearings (AMB's) have been designed to overcome the deficiencies of conventional rolling bearings. Their unique features such as no mechanical friction, leading to their ability to work in the air gap without lubrication and contamination, have led to an increasing interest in variety of applications under a wide range of temperature and pressure. However a problem common to all rotating mechanical systems, including AMB systems is the synchronous vibration caused by mass unbalance. The occurrence of such disturbances constrains the use of them in high speed applications such as aerospace industry. In general this disturbance can be modeled as combination of sinusoidal signals because of rotational vibrations. The conventional methods for disturbance rejection are mostly based on auto-balancing methods, including Filter based methods and Model based techniques. However non of these methods reduces the vibration on the rotor yet preserve desired bandwidth and frequency response.

To overcome the problems caused by conventional auto-balancing methods we propose a method known as Generalized Adaptive Forced Balancing for estimation and rejection of unknown sinusoidal disturbances. The proposed method utilizes the Adaptive Forced Balancing method along with a frequency tracker. The frequency estimation is accomplished using Adaptive Observers and Proportional Integral Adaptive Observers.

Along with application of Adaptive Observers in frequency estimation we introduced PI adaptive observers (PIAO) to facilitate the estimation process for the Linear Time Invariant systems with unknown parameters and states. The proposed PIAO method is also applied for a class of nonlinear systems to estimate the corresponding unknown parameters.

The simulations are carried out in order to estimate and reject the sinusoidal disturbances with unknown frequency and amplitude. The results indicate the proper estimation and rejection of the disturbance in an AMB system. Also the simulations on PI adaptive observers are implemented and their superior behavior are demonstrated in comparison to adaptive observers.

Thesis Committee:
Prof. B. Shafai (Advisor)
Prof. M. Sznaier
Prof. R. Sipahi