Spectrum shape based roller bearing fault detection

Abstract

This paper describes a combination of numerical methods of automated fault detection in rolling bearings, even when there is a limited knowledge about inspected bearings and their characteristic frequencies in particular. This approach approaches the problem using several rolling bearing diagnostic methods which work simultaneously and complement each other. These include the well-known Hilbert-based envelope, Digital and Continuous Wavelet Transforms, Fourier Transform and Fuzzy Logic. Several heuristic techniques are proposed, such as Outer Envelope Filtering, Pulse Train Scanning and Spectrum Shape Analysis. Combination of partial results obtained by using different techniques allows for estimating the probabilities of different fault types (outer-race, inner-race and roller).