對現代控制工程師而言，設計一個同時能滿足多種性能規格的控制系統，如:系統的穩定性、動態響應、最小特徵值靈敏度、特徵模子去耦合等，是非常重要的課題。因此僅用特徵值設計方法己不符合現代控制系統設計技術的需求。所以特徵結構的設計方法才會在最近幾十年來極受重視和關注。 近年來，為改善全狀態特徵結構設計方法之無法容易獲得非奇異模式矩陣的共同缺點，於是順序式的特徵結構設計方法被提出來。所謂的順序式的特徵結構設計方法，即一次僅從開迴路系統中設計一個或一群特徵結構到閉迴路系統。 在本論文中，我們透過狀態迴授控制，提出另一種順序式的特徵結構設計方式。其優點為可將原來的開迴路系統以對角化形式分解成較簡單的設計結構形式，且可容易的設計出非奇異的模式矩陣。而本篇所提出的順序特徵結構方法也成功的被應用在特徵模子去耦合和最小化特徵值靈敏度上。 The multi-objective control system design, such as a control system having simultaneously different performance criteria; for instance, system stability, dynamic response, minimum eigenvalue sensitivity, and mode decoupling, becomes very important tasks for the modern control engineers. Hence, eigenvalue assignment methods have been not satisfied for recent control system design techniques. Therefore, the eigenstructure assignment methods have received considerable attentions and great fruits in last decades. In recent year, a sequential eigenstructure assignment method, which can improve the common disadvantage of the entire eigenstructure assignment method, is proposed. In the algorithm of the sequential eigenstructure assignment method, only one or a group of eigenvectors are assigned at one time. It is easier to achieve the invertible modal matrix in the sequential eigenstructure assignment method than that in the entire eigenstructure assignment method. In this thesis, we propose an alternative scheme for the sequential eigenstructure assignment via linear state-variable feedback, which has the simple structure and time-saving decomposition for the original system matrix. And it has successful application to minimum eigenvalue sensitivity and mode decoupling for linear multi-input systems.
