Proposal of data mining process for tool catalog data introducing machine learning
-
- SAKUMA Taishi
- Graduate School of Science and Engineering, Doshisha University
-
- ASAKURA Akihito
- Graduate School of Science and Engineering, Doshisha University
-
- YAMADA Kotaro
- Graduate School of Science and Engineering, Doshisha University
-
- HIROGAKI Toshiki
- Graduate School of Science and Engineering, Doshisha University
-
- AOYAMA Eiichi
- Graduate School of Science and Engineering, Doshisha University
-
- KODAMA Hiroyuki
- Faculty of Engineering, Okayama University
Bibliographic Information
- Other Title
-
- 機械学習を導入した工具カタログのデータマイニングプロセスの提案
Abstract
<p>We attempt to construct a novel technology development utilizing big data such as Deep Learning in the manufacturing industry. Especially, we look at the data mining method and the tool catalog as a useful big data base which is updated by tool makers because it is easy for CAD/CAM engineers and machine tool operators to obtain it in the manufacturing fields. In the present report, we proposed the visualization and consideration of cutting condition determination process based on a decision tree method which is one type of statistical analysis method for radius-endmill data base. We also developed a cutting condition prediction system with a random forest which is a type of machine learning method applying a decision tree. Moreover, we performed a case study in endmilling under deriving cutting conditions by the proposed method, which is an unknown and expanded cutting condition based on tool catalog data base. As a result, it is demonstrated that the support based on machine learning is found to be effective to select a cutting condition including an unknown cutting condition in tool catalog data base.</p>
Journal
-
- Transactions of the JSME (in Japanese)
-
Transactions of the JSME (in Japanese) 85 (877), 19-00215-19-00215, 2019
The Japan Society of Mechanical Engineers
- Tweet
Details 詳細情報について
-
- CRID
- 1390282752331276160
-
- NII Article ID
- 130007711794
-
- ISSN
- 21879761
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
- Crossref
- CiNii Articles
-
- Abstract License Flag
- Disallowed