Gear cracks are typically difficult to diagnose with sufficient time before catastrophic damage occurs. Significant damage must be present before algorithms appear to be able to detect the damage. Frequently there are multiple gear meshes on a single shaft. Since they are all synchronous with the shaft frequency, the commonly used synchronous averaging technique is ineffective in removing other gear mesh effects. Carefully applying a filter to these extraneous gear mesh frequencies can reduce the overall vibration signal and increase the accuracy of commonly used vibration metrics. The vibration signals from three seeded fault tests were analyzed using this filtering procedure. Both the filtered and unfiltered vibration signals were then analyzed using commonly used fault detection metrics and compared. The tests were conducted on aerospace quality spur gears in a test rig. The tests were conducted at speeds ranging from 2500 to 5000 revolutions per minute and torques from 184 to 228 percent of design load. The inability to detect these cracks with high confidence results from the high loading which is causing fast fracture as opposed to stable crack growth. The results indicate that these techniques do not currently produce an indication of damage that significantly exceeds experimental scatter.

The first book on Prognostics and Health Management of Electronics Recently, the field of prognostics for electronic products has received increased attention due to the potential to provide early warning of system failures, forecast maintenance as needed, and reduce life cycle costs. In response to the subject's growing interest among industry, government, and academic professionals, this book provides a road map to the current challenges and opportunities for research and development in Prognostics and Health Management (PHM). The book begins with a review of PHM and the techniques being developed to enable a prognostics approach for electronic products and systems. building on this foundation, the book then presents the state of the art in sensor systems for in-situ health and usage monitoring. Next, it discusses the various models and algorithms that can be utilized in PHM. Finally, it concludes with a discussion of the opportunities in future research. Readers can use the information in this book to: Detect and isolate faults Reduce the occurrence of No Fault Found (NFF) Provide advanced warning of system failures Enable condition-based (predictive) maintenance Obtain knowledge of load history for future design, qualification, and root cause analysis Increase system availability through an extension of maintenance cycles and/or timely repair actions Subtract life cycle costs of equipment from reduction in inspection costs, down time, and inventory Prognostics and Health Management of Electronics is an indispensable reference for electrical engineers in manufacturing, systems maintenance, and management, as well as design engineers in all areas of electronics.

This book gives a complete presentatin of the basic essentials of machinery prognostics and prognosis oriented maintenance management, and takes a look at the cutting-edge discipline of intelligent failure prognosis technologies for condition-based maintenance. Presents an introduction to advanced maintenance systems, and discusses the key technologies for advanced maintenance by providing readers with up-to-date technologies Offers practical case studies on performance evaluation and fault diagnosis technology, fault prognosis and remaining useful life prediction and maintenance scheduling, enhancing the understanding of these technologies Pulls togeter recent developments and varying methods into one volume, complemented by practical examples to provide a complete reference

Gear cracks are typically difficult to diagnose with sufficient time before catastrophic damage occurs. Significant damage must be present before algorithms appear to be able to detect the damage. Frequently there are multiple gear meshes on a single shaft. Since they are all synchronous with the shaft frequency, the commonly used synchronous averaging technique is ineffective in removing other gear mesh effects. Carefully applying a filter to these extraneous gear mesh frequencies can reduce the overall vibration signal and increase the accuracy of commonly used vibration metrics. The vibration signals from three seeded fault tests were analyzed using this filtering procedure. Both the filtered and unfiltered vibration signals were then analyzed using commonly used fault detection metrics and compared. The tests were conducted on aerospace quality spur gears in a test rig. The tests were conducted at speeds ranging from 2500 to 5000 revolutions per minute and torques from 184 to 228 percent of design load. The inability to detect these cracks with high confidence results from the high loading which is causing fast fracture as opposed to stable crack growth. The results indicate that these techniques do not currently produce an indication of damage that significantly exceeds experimental scatter. Decker, Harry J. Glenn Research Center NASA/TM-2003-212312, NAS 1.15:212312, E-13899, ARL-TR-2933

NA4 is a vibration diagnostic parameter, developed by researchers at NASA Glenn Research Center, for health monitoring of gears in helicopter transmissions. NA4 reacts to the onset of gear pitting damage and continues to react to the damage as it spreads. This research also indicates NA4 reacts similarly to load variations. The sensitivity of NA4 to load changes will substantially affect its performance on a helicopter gearbox that experiences continuously changing load throughout its flight regimes. The parameter NA4 has been used to monitor gear fatigue tests at constant load. At constant load, NA4 effectively detects the onset of pitting damage and tracks damage severity. Previous research also shows that NA4 reacts to changes in load applied to the gears in the same way it reacts to the onset of pitting damage. The method used to calculate NA4 was modified to minimize these load effects. The modified NA4 parameter was applied to four sets of experimental data. Results indicate the modified NA4 is no longer sensitive to load changes, but remains sensitive to pitting damage.

Several interpolation techniques were investigated to determine their effect on time synchronous averaging of gear vibration signals and also the effects on standard health monitoring diagnostic parameters. The data was also digitally resampled to determine the effect of lower acquisition rates. The analysis used previously recorded vibration data taken during Health and Usage Monitoring gear testing at the NASA Glenn Research Center. The gear testing monitored the development of surface pitting fatigue on aerospace quality spur gears. Linear, cubic and spline interpolation methods were investigated. Comparisons between the resultant averages show that while there are differences in the resultant time synchronous averages, the differences are not obvious. The diagnostic parameters tested were FM4 and NA4. There are significant differences in the percent deviation curves which imply that the magnitudes of the errors increase as the sample rate decreases.