Rolling bearing monitoring algorithm by wavelet scaling comparison

Abstract

A rolling bearing is the most common element of any rotary mechanism design and, at the same time, the most vulnerable element that determines the operability and the durability of equipment. The purpose of the work is to develop the method of bearing condition control by vibration parameters on the basis of scale graph comparison of a continuous wavelet analysis.The goal is achieved by the solution of the following tasks: to determine informative criteria that allow to detect the bearing defects by the analysis of its vibration; to develop the algorithm of bearing defect detection by the parameters of their vibrations, which make it possible to control products in automatic mode.They performed the simulation of whole and defective rolling bearing signals by the synthesis of frequency components characteristic for defects.The correlation coefficient of spectra and the rank correlation of Spearman's spectra were used in the work for spectrum comparison. The comparison of the scaling graphs obtained by continuous wavelet transformation was carried out using PSNR metric.The approach is suggested in order to separate the bearings into "defect-free" or "defective" S. E. Rinatovich et al. J Fundam Appl Sci. 2017, 9(2S), 957-971 958 estimate p is calculated; the variance estimate S is calculated as the median of absolute deviations with respect to a position estimate; a confidence interval is developed for a given level of significance  . The bearings with comparison results that fall within the confidence interval are considered defect-free ones, and the ones that didn't fall within this interval are considered as defective.