A preliminary estimation of analysis methods of vibration signals at fault diagnosis in ball bearings

In this paper, a preliminary estimation of the most common analysis methods of the vibration signals of a ball bearing is tried. The tested methods are the typical statistic analysis method, the Fourier transform, the frequencies spectrum analysis and the wavelet method. With the statistical signal analysis, the quantitatively differentiation (but not qualitatively) of the measured signals in the level of vibration is easily detected, which is potentially owed in some fault bearings. With the Fourier transform detected, the frequencies of damage are presented during the vibration of a bearing. The use of spectral of frequencies of signals as well as energy that is contained in them assists to detect and/or confirm the frequencies that were already detected with the Fourier transformation. The analysis with use of continuous transformation with wavelets seems to be more promising than the discrete one, but a further experimentation is needed with various loading conditions and other type of bearings before a safe result is concluded.