應用模糊類神經網路於穩健比例積分控制器設計

Abstract

在物理世界中，任何系統模型皆含有不確定因素導致系統模型產生變化及誤差，例如系統模型估算不夠準確、系統模型過份簡化、操作過程產生誤差或是外在環境改變產生系統模型變化等因素，皆會使系統產生變化。本論文將固定參數系統設計提昇至變動參數系統設計，主要利用Kharitonov定理所衍生極值系統（extremal system）觀念在奈氏圖（Nyquist plots）、尼可圖（Nichols chart）、波德圖（Bode plots）上討論不確定性穩定度分析（analysis of uncertainly stability），進而算出基於增益邊界（GM）及相位邊界（PM）為基礎之實用性高的比例積分（PI）控制器，最後將結合模糊類神經網路（FNN）方法，找到符合我們所要頻域規格（GM、PM）的PI穩健控制器。

In the physical world, the uncertainties encountered in control system are both in the environment and within the system. These uncertainties can occur, for example, the simplified due to models or the equipment is exposed to the environment of temperature and pressure. So, we can not just design a controller for fixed plant transfer functions in classical control system.. In the parametric interval system, we still used Nyquist plots, Nichols chart , Bode plots to discuss robust stability and design robust controller. In the thesis , we propose a PI tuning method using fuzzy neural network based on gain and phase margin（FNGP） specification. We use the fuzzy neural networks to determine the parameters of PI controllers. The FNGP is able to automatically tune the PI controllers parameters with different gain and phase margin specifications, so that neither numerical methods nor graphical methods have to be used, This makes it easy to tune the controller parameters to have the specified robustness and performance.