Centralized optimal position control for active magnetic bearings: comparison with decentralized control

This paper discusses a closed-loop centralized control for a multi-input multi-output active magnetic bearing system. A genetic algorithm for design and a credible comparison of different controller structures are proposed. The evaluation of the design trade-offs of linear-quadratic and loop transfer recovery controllers are studied. The model-based controllers are compared with the classical PID controller and the cascaded PI/PD controller. The properties of the tested control configurations are examined using maximum singular values of the output sensitivity function of the closed-loop system and the tolerated disturbance at the input of the plant. Furthermore, indices such as measured peak output sensitivity, singular values of the uncertain control system, responses to the step reference position, step disturbance, and impulse force disturbance are examined. A good agreement between the simulation and the experimental results from the test-rig is shown.