딥러닝을 이용한 2-D 직조 탄소섬유강화 복합재료의 응력-변형률 선도 예측Prediction of 2-D woven CFRP stress-strain curves using deep neural networks
- Other Titles
- Prediction of 2-D woven CFRP stress-strain curves using deep neural networks
- Authors
- 정성우; 김덕중; 남병군; 김학성
- Issue Date
- Jun-2021
- Publisher
- 대한기계학회
- Citation
- 대한기계학회 재료 및 파괴부문 2021년도 춘계학술대회 논문집, pp.105 - 105
- Indexed
- OTHER
- Journal Title
- 대한기계학회 재료 및 파괴부문 2021년도 춘계학술대회 논문집
- Start Page
- 105
- End Page
- 105
- URI
- https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/191345
- Abstract
- Woven carbon fiber-reinforced plastics (CFRP) has been widely used in automotive industries due to its superior strength and stiffness to weight ratio. However, it is hard to predict mechanical behavior of woven CFRP because of its complex structure, which increases the engineering cost to design it. In this study, mechanical behavior of woven CFRP was predicted by deep neural network (DNN) models. Dataset was made by multiscale simulation depending on structures (yarn width, space, height and volume fraction) and mechanical properties of fiber and matrix. Structures and mechanical properties were used as input of DNN, and stressstrain behavior was used as output of DNN. For cost-effective train, the number of points on stress-strain curve was reduced by principal component analysis (PCA) to reduce a model size and computing time. The range of mechanical properties of constituents
and dimensions for geometric modeling is selected in consideration of that of commercial products. Finally, the prediction of stressstrain curve by DNN only takes 2 seconds which is far shorter than conventional FEM based simulation, which accompanying
modeling, meshing, calculation, and etc. The predicted results are presented and are in good agreement with experimental and simulation result.
- Files in This Item
-
Go to Link
- Appears in
Collections - 서울 공과대학 > 서울 기계공학부 > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/191345)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.