Feature tracking and prediction for gear tooth breakage prognostics

摘要

Condition-based maintenance systems monitor the operation of a mechanical component to provide the operator with an accurate assessment of the system's current condition. For real-time assessment, algorithms for diagnostics and prognostics are needed. Prognostic capabilities for gear tooth failures within an industrial gearbox were evaluated. Algorithms for extracting signal processing features from the raw vibration time signature data were investigated. Three features for each of three "events" were selected based upon their prognostic capability.