類神經網路在系統偵錯與診斷之研究

The study of system fault detection and diagnosis

吳育嘉(Yu-Chia Wu)

指導教授 ： 陳以明

元智大學/工程學院/工業工程與管理學系/碩士(2004年)

摘要

飛機次系統(如致動器或感測器故障)失效將肇致飛機無可挽回的損失，然而飛機駕駛以其經驗補償修正其失效發生，尤其在緊急狀況時更需要有電腦輔助飛機失效尌的偵錯與診斷。本研究係使用類神經網路快速偵測致動器內的感測器失效及失麥位置之辨識，本文首先使用BEAVER型機模擬產一飛行數據，提供多層式類神經網路之偵錯與RBF網路之診斷。經過不同模擬飛行數據訓練，証明多層式類神經網路與RBF網路可對不同致動器內之感測器失效時，可有效達到偵錯與診斷之目的。

關鍵字

偵錯與診斷

並列摘要

Aircraft subsystem failures (e.g. actuator or sensor failures) can cause catastropic failures that can lead to loss of the aircraft. While experienced pilots can often compensate for failure, in certain emergency situations there is the need for computer-assisted detection and diagnosis the failures to save the aircraft. In this thesis, we show that the neural networks can quickly detect the faulty sensor and identify the failure location. After establishing, the flight simulation test bed for the BEAVER aircraft. We introduce the multi-layer neural network and RBF network for the fault detection and diagnosis. Finally, we investigate the performance of the detection and diagnosis for various failure conditions on the BEAVER aircraft.

並列關鍵字

detection and diagnosis