Study of Inspection Technology for Gear Surface using Self-organizing Map
|
Accession number;05A0305024
|
| Title;Study of Inspection Technology for Gear Surface using Self-organizing Map |
| Author;
EGUCHI JUNJI
(Honda R & D. Co., Ltd., Tochigi Labs.)
MURAKAMI MANABU
(Honda R & D. Co., Ltd., Tochigi Labs.)
UCHIDA TAKANAO
(Honda R & D. Co., Ltd., Tochigi Labs.)
FUJIWARA KIN'YA
(Honda R & D. Co., Ltd., Tochigi Labs.)
|
Journal Title;Honda R&D Tech Rev
|
Journal Code:L0353A
|
ISSN:0915-3918
|
|
VOL.17;NO.1;PAGE.80-87(2005)
|
| Figure&Table&Reference;FIG.15, TBL.1, REF.4 |
| Pub. Country;Japan |
| Language;Japanese |
| Abstract;Research was carried out on inspection technology for gear surface imperfection inspection that has the same degree of accuracy as that of a human inspector. Regarding the surface of the tooth which is the object of inspection, over-detection resulting from the extraction of uneven brightness other than that due to imperfections was an issue. For this reason, the features of imperfections that permit imperfections to be discriminated from non-imperfections were set based on visual observation. In addition, a self-organizing map which is a kind of neural network was employed as a method whereby these features are used to identify imperfections in the same way as a human inspector. By using this map to teach the inspection system the features of multiple imperfections and also uneven brightness and dirt on the surface of the gear, it was possible to recognize imperfections to the same degree of accuracy as that of a human inspector. Technology constructed using this imperfection recognition method enabled the work of visual inspection to be reduced to about 10% of that necessitated previously. (author abst.) |
|
|
|
Related Articles;
|
|
