Robust H-infinity Control Of Active Magnetic Bearing System

The primary focus of this dissertation is the practical application of robust H-infinity control theory on the Active Magnetic Bearings (AMB) experimental platform. Initially every segment of the AMB experimental platform including rotor dynamics, sensors, amplifiers and controller structure is analyzed, then an accurate five-axes AMB mathematical modeling is obtained as a result which would be great useful for centralized control. And a mathematical modeling for decentralized control is gained by imperative decoupling.For the parameter uncertainties in AMB system, H-infinity controller of single axes is worked out by solving two-Riccati equations based on parameter uncertainty H-infinity control theory using system structure design method and decentralized control strategy. What's more, in order to restrain the unmodelled dynamical uncertainties in AMB system, H-infinity controllers in thrust and radial axes are further designed base on H-infinity mixed-sensitivity control theory, during which terms of inner connections between weighting function and subsequent controller are put forward, and these rules contribute to an effective way on how to select weighting functions according to closed-loop performance. Then both of the methods are qualified by system simulation on MATLAB/SIMULINK, of which the stimulant levitation results prove that H-infinity controller performs well in signal track and noise rejection theoretically.After bilinear transformation, H-infinity controllers are made up into C program in AMB experiment platform based on PC real-time control system. Finally, the laboratory rotor is suspended successfully both in static and high rotation speed state with its displacement error restrained within±0.5μm in radial,±3.6μm in thrust axes when static and±15μm in radial,±3.6μm in thrust axes when the rotation speed is 30000 r/min. This very experiment result indicates that the controllers designed have a good robust stability and robust performance upon system's parameter uncertainties and dynamic uncertainties.