用順滑模態控制與乘積網路來作非線性動態系統的穩健控制

Abstract

最近幾年，非線性動態系統的穩健控制已有許多努力與成果。本論文提出 一種由順滑模態控制與乘積網路結合而成的穩健控制。基本上，順滑模態 控制主要觀念是欲將狀態軌跡導引至一特定平面，而後保持狀態軌跡在此 平面上。根據此法則，再加上一種特定的類神經網路─乘積網路，可以設 計出一種穩健控制器。本方法的特點包括。減少順滑模態控制中的切跳問 題、提供一個系統化的控制器設計過程、和有效抗拒系統參數變化及外來 干擾。本文舉兩個例子來驗證模擬此方法的優點。 Robust control of nonlinear dynamic systems has been studied for years. In this dissertation, a new robust controller design of nonlinear dynamic systems is proposed by combining with sliding mode control and productive networks. Essentially, the sliding mode control uses discontinuous control action to drive state trajectories toward a specific hyperplane. This principle provides a guideline to design a robust controller. Productive networks, which is a special type of artificial neural networks, is then used to tackle the drawbacks of SMC. Attractive features of the proposed method include a systematic procedure of controller design, a reduction in chattering, robustness against model uncertainties and external disturbances. Two numerical examples are given to demonstrate the effectiveness of the proposed method.