循跡飛行,是評估無人飛機飛行性能非常重要的項目之一。本論文從單一航道之循跡飛行開始,根據人為駕駛經驗以及模糊控制理論,撰寫Matlab程式,設計模糊導引控制器。並藉由Matlab工具盒ODE45之協助,模擬GPS導航的回授信號,終於獲致合適的單一航道循跡模擬飛行效果。而後,根據上述單一航道之循跡模擬飛行所獲得的結果,經過適當的參數修正,擴及任意矩形四航點以及任意角度航點之循跡模擬飛行,一一皆能應付裕如。最後,本研究引用實數編碼的基因演算法,調適模糊歸屬函數的底集區間;利用隨機產生的實數編碼染色體,經由複製、交配、突變以及菁英策略(elitism)(保留最佳基因法)等過程,憑藉性能指標(performance index)的導引,調適UAV模糊導引控制器的輸入及輸出歸屬函數,進而獲得更滿意之循跡模擬飛行結果。
The path tracking of autonomous flying is an important performance index for UAV (unmanned air vehicle). In this paper, we have developed some virtual fuzzy logic controllers to simulate various 2D flying combinations of linear courses successfully by means of Matlab codes and its toolbox ODE45 step by step. These combinations include any straight courses, any rectangle-like four linear courses and combinations of arbitrary straight flying paths. Furthermore we utilize real coded GA (Genetic Algorithm) to tune the ranges of membership functions, and then promote the performance of above mentioned fuzzy logic controllers.