A machine-learning-enhanced hierarchical multiscale method for bridging from molecular dynamics to continua
- PDF / 2,112,727 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 42 Downloads / 159 Views
(0123456789().,-volV)(0123456789(). ,- volV)
EXTREME LEARNING MACHINE AND DEEP LEARNING NETWORKS
A machine-learning-enhanced hierarchical multiscale method for bridging from molecular dynamics to continua Shaoping Xiao1 • Renjie Hu2 • Zhen Li2 • Siamak Attarian1 • Kaj-Mikael Bjo¨rk3,4 • Amaury Lendasse3,5 Received: 1 January 2019 / Accepted: 29 August 2019 Springer-Verlag London Ltd., part of Springer Nature 2019
Abstract In the community of computational materials science, one of the challenges in hierarchical multiscale modeling is information-passing from one scale to another, especially from the molecular model to the continuum model. A machinelearning-enhanced approach, proposed in this paper, provides an alternative solution. In the developed hierarchical multiscale method, molecular dynamics simulations in the molecular model are conducted first to generate a dataset, which represents physical phenomena at the nanoscale. The dataset is then used to train a material failure/defect classification model and stress regression models. Finally, the well-trained models are implemented in the continuum model to study the mechanical behaviors of materials at the macroscale. Multiscale modeling and simulation of a molecule chain and an aluminum crystalline solid are presented as the applications of the proposed method. In addition to support vector machines, extreme learning machines with single-layer neural networks are employed due to their computational efficiency. Keywords Extreme learning machine Hierarchical multiscale method Molecular model Continuum model
1 Introduction & Renjie Hu [email protected] Shaoping Xiao [email protected] Zhen Li [email protected] Siamak Attarian [email protected] Kaj-Mikael Bjo¨rk [email protected] Amaury Lendasse [email protected] 1
Department of Mechanical Engineering, The University of Iowa, Iowa City, USA
2
Department of Industrial and System Engineering, The University of Iowa, Iowa City, USA
3
Arcada University of Applied Sciences, Helsinki, Finland
4
Hanken School of Economics, Helsinki, Finland
5
Information and Logistics Technology Department, The University of Houston, Houston, USA
To accelerate and foster the maturation of technology in designing novel engineering materials and devices, numerical methods [41] play an important role in exploiting new engineering design procedures. Recent developments in nanotechnology demand that molecular building blocks complement and enhance new engineering techniques at the macroscale [69, 71, 72]. Therefore, an aggressive development of new computational methods, including multiscale methods [42], is required to address complex physical phenomena at various length and time scales for the integrated design of multiscale, multifunctional materials and products [48]. Multiscale methods have been categorized into two classes: concurrent and hierarchical multiscale methods. Concurrent multiscale methods [51] employ an appropriate model to couple multiple length/time scales so that simulat
Data Loading...
