Robust control of active magnetic bearing rotors and bearingless machines

Abstract

This doctoral dissertation addresses the challenge of designing robust, scalable, and experimentally validated control strategies for high-speed bearingless machines and active magnetic bearing (AMB) systems operating under industrially relevant conditions. The work focuses on managing cross-coupling effects, structural flexibility, and operational uncertainties in applications such as kinetic compressors and precision linear motion platforms. The proposed methodology integrates finite element modeling (FEM), modal reduction, and signal-based H∞ and LPV H∞ control synthesis. The mechanical models incorporate FEM-based lookup tables for electromagnetic force and inductance, dependent on angular displacement, current, and air gap eccentricity. The control design process includes frequency-domain performance shaping and worst-case disturbance rejection, with experimental validation on several systems including a three-AMB drivetrain, bearingless compressors, and a magnetically levitated linear motion platform.

The findings show that robust H∞ controllers synthesized at zero speed remain effective across a range of speeds. FEM-based models achieve close agreement with analytically derived force equations, supporting effective decoupling of suspension and motoring dynamics. Linear parameter-varying (LPV) H∞ controllers offer advantages in frequencydomain shaping but yield comparable time-domain performance for compact rotors. In high-temperature heat pump applications, bearingless motors outperform AMB-based solutions in terms of force density and energy efficiency. These contributions bridge the gap between theoretical robust control synthesis and practical deployment in levitated machines, providing validated tools and design strategies for both next-generation highspeed rotating and linear machines.