| Summary: | 博士 === 國立臺灣大學 === 電機工程學研究所 === 101 === This dissertation studies 1) the controllability and observability of discrete-time linear time-delay systems and 2) the predictive tracking control of a class of discrete-time linear systems with an input delay. In the first part, focus is on the controllability and observability of discrete-time linear time-delay systems, so that the two properties can play a more fundamental role in system analysis before controller and observer design is engaged. Complete definitions of controllability and observa- bility, which imply the stabilizability and detectability, respectively, and determine the feasibility of eigenvalue assignment, are proposed for systems with delays in both state variables and input/output signals. Necessary and sufficient criteria are developed to check the controllability and observability efficiently. In the second part, the robust trajectory tracking control problem of discrete-time networked control systems with time-varying input delay and a time-invariant input delay are handled, based on model-based predictor and observer-based predictor, respectively. The dynamic predictive feedback linearization controller is adopted to compensate the system dynamics and input delay, so that the system output can perfectly track the desired trajectory when uncertainties are absent. Tracking errors caused by the time-varying parameter uncertainties are suppressed as much as possible. Finally, the usefulness and effectiveness of the results from both parts are illustrated by numerical examples.
|
|---|
