지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학위논문
- 저자정보
- 지도교수
- 임재혁
- 발행연도
- 2022
- 저작권
- 전북대학교 논문은 저작권에 의해 보호받습니다.
이용수8
이 논문의 연구 히스토리 (3)
- 이 논문의 후속연구가 궁금하신가요?
- 연관 학술논문 또는 학술발표를 통해 보다 발전된 연구결과를 확인하실 수 있습니다.
- 이 논문의 연구 히스토리 확인하기
초록· 키워드
오류제보하기
In this work, we propose a machine learning modeling approach using the composite material microstructure images to predict the transverse mechanical behaviors of unidirectional composites. For this prediction, representative volume element (RVE) samples were generated for each fiber volume fraction (Vf) of 40%, 50%, and 60% by the random sequential expansion algorithm (RSE) algorithm. Subsequently, the transverse mechanical behaviors were obtained by the finite element method (FEM)-based transverse tensile analysis. Then, the machine learning model was developed to predict the transverse mechanical behaviors based on microstructure using RVEs for each Vf. To confirm the performance of the proposed machine learning model, we performed transverse mechanical behaviors on various test datasets. Prediction accuracy was verified in terms of the prediction performance indexes. The prediction results were in good agreement with the test datasets, and the transverse mechanical behaviors were quickly predicted for any microstructure. This confirmed that the proposed machine learning model is fairly simple and powerful and can efficiently provide correlations between the microstructure and mechanical behaviors of composite materials. In addition, it is concluded that this model can easily generalize the relationship between the microstructure and material properties of UD composite materials with the stress-strain curves encompassing the entire material response. Finally, this approach was extended as a way to improve the neural network predictive power of DNNs through the data sampling.
목차
Chapter 1. Introduction 11.1 Research background 11.2 Research trends 41.3 Research objectives 9Chapter 2. Generation and Evaluation of RVEs 112.1 Generation of RVEs using RSE algorithm 112.2 Evaluation of statistical spatial features for RVEs 16Chapter 3. Numerical simulation of the transverse tensile behavior 263.1 Generation of FE model 263.2 Validation of FE model 373.3 Evaluation of transverse mechanical behavior 46Chapter 4. Machine learning and deep learning 504.1 Deep neural network (DNN) 594.2 Convolutional neural network (CNN) 67Chapter 5. Prediction results by machine learning and deep learning 795.1 Hyper-parameter optimization for DNN 795.2 Evaluation of the performance for DNN 865.3 Evaluation of the performance for CNN 905.4 Comparison of the performance for DNN and CNN 94Chapter 6. Improvement of the performance of DNN by using data sampling 966.1 Review of the performance for DNN 966.2 Data sampling 996.3 Evaluation of the performance for DNN 102Chapter 7. Conclusion 104References 108국 문 초 록 125
최근 본 자료
댓글(0)
0