Conventional EMS system with state or output feedback control could show unsatisfactory performance or become unstable under unexpectedly big air-gap disturbance and breaks down when the actuator or sensor fails.
This thesis presents a robust and reliable LMI-based $H_\infty$ gain scheduled controller using time-delay-based disturbance estimator and sliding-mode observer for the EMS system in such cases. It estimates the scheduling variables and the states from system responses and control input to cancel the unexpected disturbances and directly modifies the system behavior rather then merely adjusts feedback gains or identifies system parameters. The proposed controller linearizes the system by gain scheduling and the linearized closed-loop system is made asymptotically stable with $H_\infty$ - norm bound $\gamma > 0$ via proper state or output feedback. The estimated disturbance is used as scheduling variables. The gains are modified by linear matrix inequalities. The $H_\infty$ performance of the proposed controller is bounded within $H_\infty$ - norm bound $\gamma$. It is shown that excellent system performance can be obtained with the use of time-delay-based disturbance estimator and sliding-mode observer, and that chattering and estimation error of the sliding-mode observer could be reduced and effectively eliminated.
The mathematical model of the EMS system was derived in a rigorous framework including all possible dynamics which influence the system stability. A single-magnet suspension system with nonlinear equations of motion was simulated to verify the proposed LMI-based $H_\infty$ gain scheduled controller when the actuator or sensor fails under normal operating conditions. Simulations for the time-delay-based disturbance estimator and sliding-mode observer were also carried out for the single- and double-magnet suspension systems. Results show that the proposed controller worked quite well.
A full scaled double-magnet suspension syst...