Detecting mechanical failures inducing periodical shocks by wavelet multiresolution analysis. Application to rolling bearings faults diagnosis

Abstract

The aim of this article is to show the interest of the wavelet multiresolution analysis within the detection of mechanical faults inducing periodical impulsive forces. To adapt it for this purpose, several of its analysis parameters are chosen, even optimized. The kurtosis is used as an optimization, selection and evaluation criterion. Indeed, in normal conditions the distribution of the amplitudes is as Gaussian type, if a defect appears a modification is seen in the signal in form of periodical impulses, the kurtosis is very sensitive to these shocks and allows detecting the defect. First, the proposed method is applied on simulated signal, for the experimental validation several series of measurements were carried out on a 6200 ball bearings on test rig, for this purpose various defects were simulated. Finally, the industrial application is carried out on produc-tion machine.