Multiple Defect Size Estimation of Rolling Bearings using Autonomous Diagnosis and Vibrational Jerk

Abstract

Developing condition monitoring techniques for monitoring ball bearings degradation is crucial to minimize unplanned down-time and maintenance costs. A recent monitoring approach is to maximize the useful service life of bearings by delaying their replacement after detecting an incipient fatigue defect, i.e. pitting/spall or wear. A variety of techniques have been investigated and reported in the literature to monitor defect size prior to a critical threshold. These studies focus on a single defect; nevertheless, the presence of multiple defects is practically experienced and its presence may introduce inaccuracies and result in wrong defect size estimation. In this paper, an autonomous fault detection technique is used to detect defect type and location, then the vibrational jerk method is extended to quantify multiple localized bearing defects. Operational spall size estimations are achieved for two defect conditions: natural run to failure defects from a test stand at University of Agder in Norway, and artificially seeded multiple spalls provided by DLR (German Aerospace Center) in Germany.