Magnetic bearings are capable of suspending shafts rotating at high speed without mechanical contact or lubrication, providing many advantages and opportunities in industrial and medical applications. A formidable control problem is presented by their fast, interactive and unstable multivariable dynamics. This thesis provides methods for designing controllers for magnetic bearings along with their experimental evaluation.;In a novel approach, the experimental apparatus was modeled using a parameter estimation method calculated from input and output data. An optimal controller and a discrete sliding mode controller with outer-loop integration were designed. A sliding mode controller was also implemented that changed the sliding surface depending on shaft position to more aggressively achieve the performance objective of preventing shaft mechanical touchdown. Experimental testing verified the tuning goal and showed that the sliding mode controllers with guaranteed robustness performed comparably in disturbance rejection to the optimal controllers. This work summarizes conditions for controller selection. |