Abstract
The wavelet transform is expected to apply to the diagnostic technology for machines, as the transform can detect the local abnormalities of the signal. Then, it is necessary to remove the unacceptable noise in the wavelet transforms for the effective diagnosis. In this paper, we were able to detect the slight collision and the Coulomb friction by means of removing the noise from the waveforms which were limited to regular waveforms and abnormal signals were periodical. This paper presents the analytical procedure and discussions on the experimental results, and proves that the wavelet transforms is applicable to the diagnostic technology.
