Multisensor Monitoring of Gear Tooth Fatigue for Predictive Diagnostics
Abstract
Successful machine diagnostics is critically dependent on the collection and processing of prognostic features that relate back to failure precursors. Since the significance of gear tooth failure is recognized as a critical component to the overall condition of drive train mechanical systems, single gear tooth failure has been examined. By employing a special jig to orient and constrain the gear samples, Hertzian loading was applied along a single contact line on the gear tooth to simulate the conditions seen during operation. Optical, ultrasonic and mechanical sensors measured a variety of observables including load, deflection, and acoustic emission. After monitoring the fatigue test with these three noncommensurate sensors, features of the data could consistently be related to crack growth phenomena. Data collection, analysis, and interpretation are discussed for spur gear samples that show both the absence and presence of cracks and support the validity of the extracted features as failure precursors. Cyclostationary analysis, an advanced signal processing techniques, was used to promote earlier indication and sharper resolution of these measures. The results demonstrate the potential for using nontraditional sensors and techniques, which are more amenable to commercial use, for an in situ monitoring system.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1996
- Accession Number
- ADP010165
Entities
People
- Amulya K. Garga
- Clark Moose
- Grant A. Gordon
Organizations
- Pennsylvania State University