Decoupled and quantified detection of bearing spall-like faults using strain gauge array

This study proposes a strain-based methodology for decoupling, localizing, and quantitatively diagnosing bearing faults. Multiple strain gauges are circumferentially and evenly arranged on the bearing’s outer surface using precision-machined housing notches for optimal placement. Spall-type faults can be effectively classified and localized through time-domain analysis of strain-drop pulses in the acquired signals. Fault width estimation is achieved by combining three key parameters: fault transit time extracted from strain signals, relative contact velocity and rolling contact semi-width. To validate the proposed methodology, controlled experiments were conducted using bearings with artificially induced spall-like faults (0.5–2 mm in width) located on rollers, outer raceways, and inner raceways under varying rotational speeds. The elevated signal-to-noise ratio and array test of the strain signals show promising decoupled diagnostic potential for compound spall-like faults in the test bearings. Experimental results reveal that the mean estimated fault width deviates from actual measurements by less than 5% in most test cases, confirming the method’s estimation accuracy.