小波法於引擎動態研究及故障分析之初步探討

A STUDY OF WAVELETS FOR ENGINE VIBRATION AND FAULT DIAGNOSIS

本文主要目的在研究小波理論於實車引擎動態與故障時振動信號的初步探討，在本文研究結果中計有(l)FFT；(2)基於傅立葉轉換之時－頻分析與(3)小 波分析等呈現。經由實驗結果顯示小波分析可提供一種較佳的時－頻分析與時域分析可以改善(1)FFT於不穩定信號的模糊現象與(2)基於傅立葉轉換之時－頻分析中兩相鄰階次的干涉鬼影產生，並利用統計方式擷取每一頻段內的特徵值呈現出故障特徵的趨勢作為故障檢測的重要參考資料。經實驗結果分析可確實得到引擎故障現象的振動小波特徵，此特徵可應用於引擎故障的檢測與診斷上。

This paper is to study wavelet analysis used to detect and analyze engine faults. The paper includes the following sections. (1) FFT (Fast Fourier Transforms). (2) Wavelet time-frequency analysis. (3) The results of wavelet analysis. As we can see from the results, wavelet analysis can present the signals in time-frequency distribution diagrams resolved in time and frequency. Wavelet analysis provides functions to correct the following faults. (1) FFT is difficult to use, to capture nonstationary signals. (2) FFT suffers from aliasing (interfering shadow, overlap shadow) if the samples are too close together. The signals used to analyze the engines characteristic values and faults are statistical standard deviations. In a garage, a technician always uses a scan tool such as a DRB Ⅱ, a DRB Ⅲ, a FLUKE 98 etc., to access vehicle control system data, which is diagnostic trouble codes. Then the technician can follow the message to maintain the vehicle. Wavelet analysis is another way to provide in formation maintain a vehicle. It is of more academic value and is a better fault diagnostic tool for vehicle maintenance.