Fault-signature modeling and detection of inner-race bearing faults

This paper develops a fault-signature model and a fault-detection scheme for using machine vibration to detect inner-race defects. To motivate this research, it is explained and illustrated with experimental results why fault signatures from nonouter-race defects (e.g., inner-race defects) can be less salient than those from outef-race defects. Then, a signal model is presented for the production and propagation of an inner-race fault signature; this model is then used to design an inner-race fault-detection scheme. This scheme examines machine-vibration spectra for peaks with phase-coupled sidebands occurring at a spacing predicted by the model. The proficiency of this fault-detection scheme at detecting inner-race bearing faults is then experimentally verified with results from 12 bearings representing varying degrees of fault severity.