| Summary: | 碩士 === 南台科技大學 === 機械工程系 === 99 === In this paper, an envelope estimation algorithm based on the resonance modes of a bearing system is proposed to derive the envelope signals and, then, the correlation coefficient of the envelope signals could be further calculated to be a quantitative index for the bearing defect diagnosis. According to the vibration spectrum of the bearing system, the resonance frequencies in the range of its corresponding resonance modes could be derived. Under the assumption of stepwise functions for the envelope signals, the signal of bearing vibration could be decomposed into the sinusoidal function bases with fundamental frequencies at the corresponding resonance frequencies. Thus, the envelope signals could be retrieved by estimating the stepwise functions. In addition, the reconstructed signal with noise rejection could be derived from the envelope signals. According to the envelope signals, the correlation coefficient is calculated to be the quantitative index for the bearing defect diagnosis. In general, the vibration energy would be increased with the size of bearing defect and, thus, the correlation coefficient between the envelope signals would be also increased. On the other hand, the vibration of the normal bearing would be random and its vibration energy would be much weaker than that for the defect bearing. Accordingly, the correlation coefficient between the envelope signals of the normal bearing would be also much weaker than that between the envelope signals of the defect one, and it could be a quantitative index for the bearing defect diagnosis. In the simulated and experimental studies, it is found that the envelope estimation algorithm could be effectively applied in the signal processing for the bearing vibrations. In addition, the correlation coefficient could successfully be the quantitative index for the bearing defect diagnosis.
|
|---|
