Using convolutional neural networks to predict composite properties beyond the elastic limit
- PDF / 2,340,987 Bytes
- 9 Pages / 612 x 792 pts (letter) Page_size
- 17 Downloads / 171 Views
rtificial Intelligence Research Letter
Using convolutional neural networks to predict composite properties beyond the elastic limit Charles Yang*, Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA Youngsoo Kim*, and Seunghwa Ryu, Department of Mechanical Engineering & KI for the NanoCentury, Korea Advanced Institute of Science and Technology, Daejeon 34141, Republic of Korea Grace X. Gu, Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA Address all correspondence to Seunghwa Ryu at [email protected] and Grace X. Gu at [email protected] (Received 26 January 2019; accepted 4 April 2019)
Abstract Composites are ubiquitous throughout nature and often display both high strength and toughness, despite the use of simple base constituents. In the hopes of recreating the high-performance of natural composites, numerical methods such as finite element method (FEM) are often used to calculate the mechanical properties of composites. However, the vast design space of composites and computational cost of numerical methods limit the application of high-throughput computing for optimizing composite design, especially when considering the entire failure path. In this work, the authors leverage deep learning (DL) to predict material properties (stiffness, strength, and toughness) calculated by FEM, motivated by DL’s significantly faster inference speed. Results of this study demonstrate potential for DL to accelerate composite design optimization.
Introduction Composite materials, which offer a variety of advantages that cannot be gained solely with only one of their constituents, are actively used in advanced engineering applications such as lightweight structures for aerospace and automotive industries. Natural creatures also exploit composites to protect themselves from threats in a variety of environments and to sustain living conditions with the limited resources and building blocks available in nature.[1–5] Design and fabrication methods of most man-made synthetic composites have been well established owing to their relatively simple arrangements. In comparison, although extensive studies have been performed to understand and mimic natural composites, it remains a daunting task to fabricate bio-inspired structures via conventional manufacturing processes because of their complex hierarchical structure ranging from the nano- to macro-scale. Recently, the advancement of additive manufacturing has facilitated the fabrication of complex structures, and as a result, a variety of composite structures inspired by natural materials such as nacre, bone, conch-shell, and spider silk have been fabricated and tested via 3D-printing methods.[6–11] Because many natural composites, synthesized via a self-assembly process, have relatively periodic and regular arrangements, their mechanical properties can be reasonably understood by analyzing the load transfer mechanism of a
* These authors contributed equally to this work.
representative unit cell.[5,8,10,12] However,
Data Loading...
