| Autor: |
Mitsuo Ojima, Janusz Kasperkiewicz, Dariusz Alterman, Hiroshi Akita |
| Rok vydání: |
2006 |
| Předmět: |
|
| DOI: |
10.1533/9780857093080.195 |
| Popis: |
The paper is dedicated to presentation of possibilities of Machine Learning techniques, (ML), in uniaxial tension tests on brittle matrix composites. This method enables extracting in an automatic way the knowledge hidden in examples. Concrete is weak in tension but proper evaluation of its tensile strength allows better understanding of its possibilities. It is important to identify whether its tensile properties were measured without or with elimination of uncontrolled flexure which may occur in uniaxial tension tests. In the last time, many different tests on uniaxial tension with elimination of such secondary flexure have been performed in Tohoku Institute of Technology. The results have been collected in a database. The ML experiments bring an answer to a question - how can be correctly identified the two types of uniaxial tension tests - those obtained without and those obtained with a detrimental effect of the secondary flexure. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
