Low-Cost Speed-Matched Health Indicator for Bearing Fault Detection in Low-Speed Conveyor Systems

Conveyor bearing faults can cause unplanned downtime, but conventional resonance-based envelope analysis requires high-bandwidth sensing, and supervised classifiers require labelled fault data. This work proposes the Z-score Composite Health Indicator (Z-CHI), which combines speed-matched modified z-scores of root mean square (RMS), kurtosis and robust crest factor through a Euclidean norm, so that deviations in either direction will contribute to the score. The reference and detection threshold are both calibrated from healthy-bearing data only. Z-CHI was evaluated on a bench-scale conveyor rig with a 400 Hz microelectromechanical systems accelerometer (approximately USD 40 of sensing hardware) and seeded outer-race defects at 157 and 211 revolutions per minute (RPM). All six minor-condition runs exceeded the calibrated thresholds, and in a nested leave-one-out assessment, none of the twelve held-out healthy runs exceeded its threshold. A one-sided RMS comparator separated the minor defect only at 211 RPM, and a low-bandwidth envelope-power comparator at neither speed. The monotonic condition ordering at 211 RPM is a preliminary indication of potential for remaining useful life estimation.