類神經網路與田口法於引擎故障聲訊辨識之應用

• Main Authors: Pan, Kuan Cheng, 潘冠呈
• Other Authors: 林筱増
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/43044983580155652330

碩士 === 國防大學中正理工學院 === 兵器系統工程研究所 === 92 === As the improvement of modern technology, detecting of vehicle engine’s faults is achieved by computer. However, it cannot detect the noise cause abnormalities of mechanical parts. The objective of this thesis is to combine artificial neural network and Taguchi’s Method to construct an engine fault recognition system. It is expected to detect abnormalities of engine before it breakdown, and to assist mechanical technician to identify the origin of the problem. In short, this research is constructed by artificial neural network and Taguchi’s Method to identify different engine fault noises by network learning process. For noise signal normalization, OTA and TPA methods are used. In construction of artificial neural network model, number of input layer, number of hidden layer, learning rate, and learning cycle number are chosen as control factors. Taguchi’s orthogonal array is then used to conduct calculations in different levels. Signal to noise ratio, analysis of variance, and F-test are applied to analyze effect of different control factors to engine fault recognition rate, and to accomplish optimum combination for relevant control factors. First, combination of four control factors with different levels are selected randomly, the best fault recognition rates are 58.3% (OTA normalization) and 56.3% (TPM normalization). The recognition rates are raised to 66.7% (OTA) and 60.4% (TPM) by application of Taguchi’s method. Finally, with the analysis of variance and F test, the most important control factor is identified as the number of input layer. After fine adjustment of this control factor, the optimum fault recognition rate of 68.8% (OTA) and 66.7% (TPM) are achieved. The most important conclusion can be drawn form this research is that the combination of artificial neutral network and Taguchi’s Method can be applied to engine fault noise recognition problem, and the recognition rate of the artificial neutral system is raised effectively.