轉動機械之線上監測診斷系統研發

Abstract

本研究的主要目是希望研究發展一套針對轉動機械的線上監測診斷系統。工作內容將包括訊號處理及狀態推論，訊號處理部分是將訊號做運算，得到所需要的特徵值；狀態推論的部分則是由模糊類神經綱路或模糊專家系統來實現。當能取得機器所有故障狀態訊號的情形下，將機械訊號作如平均值、標準差、峰值、倍頻能量值等數位訊號處理以得系統的特徵值，然後輸入到模糊類神經綱路中讓綱路學習並做推論；若無法取得機器所有故障狀態的訊號，則將量測的加速度訊號做積分、快速傳利葉轉換（FFT）、倍頻能量值等計算，配合依照ISO 2372定出的推論法則，在該情形下進行推論。所發展各種方法皆是利用數位訊號處理器（DSP）擷取訊號並作訊號處理，以達到轉動機械的線上監測診斷的目的。

The aim of this research is to develop an on-line monitoring and diagnostic system for rotating machinery. The architecture of the systems mainly comprises a signal processing module and a state inference module. Two approaches are employed for state inference; the neural fuzzy network and the fuzzy expert system. If the information of all faults is available, the mean value, the standard deviation, the maximum, and the harmonic multiples of signals will be used as the featurese of the signal. Then, the neural fuzzy network is exploited for intelligent inference of states based on the extracted features. Otherwise, a fuzzy expert system is employed for fault inference on the basis of ISO 2372. By using a digital signal processor (DSP), the two approaches are implemented and verified in various applications of rotating machinery.