模糊类神经网路之齿轮故障诊断.PDF

先進工程學刊第二卷 第一期 41 Journal of Advanced Engineering Vol. 2, No. 1, pp. 41-47 / January 2007 模糊類神經網路之齒輪故障診斷 Gear Fault Diagnosis by Using Fuzzy Neural Network 1 *2 1 1 1 康淵 朱明輝 王俊傑 張智傑 張永鵬 1 *2 1 1 1 Yuan Kang , Ming-Huei Chu , Chun-Chieh Wang , Chih-Chieh Chang , Yeon-Pun Chang Abstract This paper applies fuzzy neural network (FNN) in the fault diagnosis for the gear-rotor system. According to the document and experiment data, the relationship between fault causes and frequency symptoms is built up as the rule of approximate reasoning diagnosis and the training data for NN. In this paper, the four typical faults for gear system will be taking into vibration measured: (1) Gear skew (2) Shaft not parallel (3) Tooth breakage (4) Gear wear. The characteristic signals are extracted by using the technique of analyzing spectrums. The symptoms classified by using membership function and trained by using FNN. The diagnosis result by using FNN compared with NN and similarity reasoning. And the accurate diagnosis will be verified by using FNN in the complex system. Keywords : fuzzy neural network, gear skew, shaft not parallel, tooth breakage, wear 摘要 本文將模糊類神經網路應用於齒輪轉子系統的故障診斷，根據既有的文獻與 實驗數據建立齒輪故障類型與頻譜特徵間的關係，作為類神經網路學習的依據。