On-line severity assessment of bearing damage via defect sensitive resonance identification and matched filtering

Abstract

A microcomputer based on-line bearing condition monitoring system was developed. Employing synchronised segmentation and parametric spectral comparisons, the system enabled on-line identification of defect sensitive resonances for an investigated bearing system at an early stage of damage. A matched filter was designed to keep track of the energy contributed by these resonances throughout the rest of bearing life. The magnitude of the energy was found to be well correlated with the development of bearing localised defects. It takes 38 sec for the identification of defect sensitive resonances and 7 sec for the matched filter to report each new assessment of bearing condition on a programmed PC-AT.