本論文研究車輛主動式懸吊系統之動態控制器設計。本文提出一套控制器設計方法，俾使車輛行駛於凹凸地面時有減震效果。 現存設計方法絕大多數採用靜態狀態回授，不同於以往，本文考慮動態輸出回授。本文提出一特殊轉換，可將動態輸出回授控制器的設計問題等價轉換為靜態狀態回授的設計問題，然後可應用現存方法完成設計。因此，車輛安全性的限制(即抓地能力與懸吊系統最大變形量)以及控制力的限制亦可整合於設計條件之中，此為歷來輸出回授設計方法難以達成的特色。另外，所研究問題可視為一窄頻雜訊之抑制問題。本文所提方法引入了內模型，並結合GKYP (generalized Kalman-Yakubovich-Popov, GKYP)引理，其能使干擾到控制輸出(即車身垂直方向的加速度)在指定頻段的H∞範數儘量小，因此可提高乘坐舒適性。 本論文所提出的設計條件皆為線性矩陣不等式(Linear Matrix Inequalities，LMIs)，故可運用現存的軟體有效求解。本文運用Matlab及Simulink進行數值模擬，設計出動態控制器，將其與未引入內模型的全頻段H∞控制器進行頻域波德圖和時域響應圖之比較，模擬結果證實了所提設計方法之可行性及優越性。 In this thesis, the dynamic controller synthesis problem of vehicle active suspension systems is investigated. A controller design method is proposed which is capable of reducing the vibration of the vehicle when driving on uneven ground. In contrast with the majority of the existing design methods which employ static state feedback, this thesis considers dynamic output feedback. A special transformation is introduced that equivalently converts the dynamic output feedback controller synthesis problem into a static state feedback design problem. Then the design can be completed by applying the available existing methods. The introduced transformation overcomes the difficulty of incorporating the constraints of road holding ability, maximum suspension deflection of the active suspension system, and actuator limitation into a output feedback design. This feature is hardly seen in the existing methods. In addition, the control problem is regarded as a narrow-band disturbance attenuation problem. To enhance ride comfort, the proposed methods introduce internal models into the design, combined with the generalized Kalman-Yakubovich-Popov (GKYP) Lemma in order to minimize the H∞ gain of the transfer function from the disturbance to the observed output (i.e., the vertical acceleration of the vehicle body) in the desirable restricted frequency range. Synthesis conditions are derived in terms of linear matrix inequalities (LMIs), which can be efficiently solved by existing LMI solvers. Numerical simulation is conducted using Matlab and Simulink. Dynamic controllers are computed by the proposed methods and compared with the (entire frequency) H∞ controller design without introducing the internal models. Simulation results demonstrate the feasibility and effectiveness of the proposed design methods in terms of lower Bode diagrams and better time domain responses.