本文是針對穿戴式下肢輔具的需要而設計的一個可攜式DSP/CPLD系統。本文整合工業控制器(LinCon8K)、數位訊號處理器 (Digital Signal Processor, DSP)、複雜型可規劃邏輯閘 (Complex Programmable Logic Device, CPLD)及Linux 作業系統以實現此一氣壓控制系統。 本文提出結合小波轉換之反覆學習控制律應用於髖關節氣壓伺服控制系統,以解決傳統控制器難以處理的非線性以及不確定性的問題,使氣壓缸追循任意軌跡之曲線。反覆式學習控制(Iterative Learning Control, ILC),是一經由反覆的控制系統過程中記錄系統所輸入、輸出的資訊,以作為下次控制力修正的參考,且反覆學習控制能達到在有限的學習次數中,將系統的跟隨誤差收斂到一有限的範圍內。但在學習過程中可能會受到外界干擾,導致學習效能不佳,因此期望利用小波轉換來處理不可學習之動態,以確保學習曲線之收斂性。
Based on the need of a wearable lower limb orthosis, this thesis is aimed to design a portable DSP/CPLD system that allows user to move freely with the orthosis. Here, we integrate an industrial controller (LinCon8K)、DSP(Digital Signal Processor)、CPLD(Complex Programmable Logic Device) and Linux operating system to realize this pneumatic control system. This thesis presents a control law that combines wavelet and ILC on the control of a pneumatic servo system. It can provide a solution to the nonlinear and uncertain problems that the traditional controller can not achieve. An Iterative Learning Control (ILC) scheme would use the information in repetitive operation and progressively improving the tracking performance. The tracking error can be converged to a small iterative learning controller would only have repeatable uncertainties. Here, wavelet transform was applied to handle the un-learnable dynamics to ensure the convergence of this iterative learning process. Thus the pneumatic cylinder can track an arbitrary trajectory profile.