• 作者： Chen, Chyi-Tsong; Hwu, Jong; Chang, Wei-Der
• 中文摘要： 本文提出一基於自調諧單神經元之非線性控制策略。所採用之單神經元控制器架構簡單，具高度非線性映射能力，配合以所衍導之參數調諧法即可隨系統變化，自適應地調整控制器參數，以使系統輸出以最小平方誤差的方式逼近所欲的系統輸出值。此控制器由於具有自調諧能力，可自動調整其函數型態，因而控制力輸出範圍可自動調整，無須額外尺度調整的工作，可避免控制力飽合問題。同時，調整參數的過程僅需系統實際輸出訊息，不須複雜的控制理論背景知識，極適合做直接式線上適應控制，其在應用上有相當的實用價值。在本文中，吾人將此控制策略應用於不穩定之非線性反應器控制上，並與多層之類神經網路控制器作深入地比較。由模擬之結果顯示，吾人所提之非線性控制策略具有簡單性及易用性，深具應用前景。
• 英文摘要： This paper develops a single neuron control system for the direct adaptive control of single-input/single-output nonlinear chemical processes. A single neuron, which possesses three adjustable shape parameters, is adopted as the process feedback controller. To enable the single neuron controller (SNC) to learn to control a changing process adaptively, a simple parameter tuning algorithm is proposed. The presented adaptive control strategy is easy-to-implement and directly operational, where no a priori knowledge of the controlled system except its output response direction is required. The effectiveness and applicability of the proposed scheme were demonstrated through the challenging problem of controlling a nonlinear and open-loop unstable CSTR. Furthermore, the performance of the SNC-based control system was compared extensively with that of using a multi-layered neural controller. Simulation results show that the proposed nonlinear control strategy appears to be a simple yet powerful approach to the direct adaptive control of nonlinear chemical processes.
• 中文關鍵字： 非線性程度控制; 單神經之控制器; 參數調諧法; 非線性連續攪拌槽反應器
• 英文關鍵字： Nonlinear Process Control; Single Neuron Controller; Parameter-Tuning Algorithm; Nonlinear Cstr