| Condition Based Maintenance--or the servicing of machinery only when it needs it instead of on a strictly scheduled basis--can increase safety and lower costs of complex mechanical systems. The key to Condition Based Maintenance is being able to determine the condition of a mechanical component while the component is in service. Vibration is a good way of getting condition information, since it is both easy to measure and closely related to mechanical conditions. Gearboxes are common mechanical components that also require much maintenance and are often involved in catastrophic failures, particularly in helicopters and other high stress, low weight systems.; This research contributes toward the goal of using vibration for Condition Based Maintenance for gearboxes in four ways: First, a custom test setup has been created for inducing a controlled failure in a gearbox while taking vibration, stiffness, and visual measurements. Vibration data can be taken autonomously over an indefinite period of time. Data has been collected on this setup during two run-to-failure tests. Second, a prototype wireless sensor particularly suited for shaft-mounted torsional vibration measurement has been produced. Third, a family of model error based algorithms has been developed for extracting gear condition information from gearshaft vibration. These algorithms perform comparably with, and in some cases better than, existing algorithms. Fourth, a bearing fault detection algorithm has been developed that is superior to existing algorithms for early detection of cage failure from vibration data.; These contributions bring Condition Based Maintenance a step closer to implementation. While not constituting a Condition Based Maintenance system in itself, the tools developed in this thesis give future designers of such systems some very effective options. |
