NASA Logo

NTRS

NTRS - NASA Technical Reports Server

Back to Results
Gear Tooth Wear Detection AlgorithmVibration-based condition indicators continue to be developed for Health Usage Monitoring of rotorcraft gearboxes. Testing performed at NASA Glenn Research Center have shown correlations between specific condition indicators and specific types of gear wear. To speed up the detection and analysis of gear teeth, an image detection program based on the Viola-Jones algorithm was trained to automatically detect spiral bevel gear wear pitting. The detector was tested using a training set of gear wear pictures and a blind set of gear wear pictures. The detector accuracy for the training set was 75 percent while the accuracy for the blind set was 15 percent. Further improvements on the accuracy of the detector are required but preliminary results have shown its ability to automatically detect gear tooth wear. The trained detector would be used to quickly evaluate a set of gear or pinion pictures for pits, spalls, or abrasive wear. The results could then be used to correlate with vibration or oil debris data. In general, the program could be retrained to detect features of interest from pictures of a component taken over a period of time.
Document ID
20170005246
Acquisition Source
Glenn Research Center
Document Type
Technical Memorandum (TM)
Authors
Delgado, Irebert R.
(NASA Glenn Research Center Cleveland, OH, United States)
Date Acquired
June 7, 2017
Publication Date
July 1, 2015
Subject Category
Aeronautics (General)
Report/Patent Number
NASA/TM-2015-218830
E-19102
Funding Number(s)
WBS: WBS 664817.02.03.02.03.02
Distribution Limits
Public
Copyright
Work of the US Gov. Public Use Permitted.
No Preview Available