1. Bearing fault detection method based on statistical analysis and KL distance
- Author
-
Emre Yengel, Arda Mollakoy, and B. Ugur Toreyin
- Subjects
Bearing (mechanical) ,Kullback–Leibler divergence ,Computer science ,business.industry ,Seal (mechanical) ,law.invention ,Hough transform ,Visual inspection ,Natural rubber ,law ,visual_art ,visual_art.visual_art_medium ,Statistical analysis ,Computer vision ,Artificial intelligence ,business - Abstract
The final step of the bearing production line constitutes the inspection of the bearing which is mostly performed by visual inspection. Three groups of bearings namely, properly assembled samples, conversely assembled rubber seal and samples where rubber seals were missing are classified using visible range images of these samples. According to the proposed method, extraction of seal regions from the bearing images using circular Hough transform is followed by a higher-order statistical analysis to finalize the classification. Experimental results show that this system may be employed as an assistive tool for bearing inspectors.
- Published
- 2016