Abstract
Topology optimization problems pose substantial requirements in computing resources, which become prohibitive in cases of large-scale design domains discretized with fine finite element meshes. A Deep Learning-assisted Topology OPtimization (DLTOP) methodology was previously developed by the authors, which employs deep learning techniques to predict the optimized system configuration, thus substantially reducing the required computational effort of the optimization algorithm and overcoming potential bottlenecks. Building upon DLTOP, this study presents a novel Deep Learning-based Model Upgrading (DLMU) scheme. The scheme utilizes reduced order (surrogate) modeling techniques, which downscale complex models while preserving their original behavioral characteristics, thereby reducing the computational demand with limited impact on accuracy. The novelty of DLMU lies in the employment of deep learning for extrapolating the results of optimized reduced order models to an optimized fully refined model of the design domain, thus achieving a remarkable reduction of the computational demand in comparison with DLTOP and other existing techniques. The effectiveness, accuracy and versatility of the novel DLMU scheme are demonstrated via its application to a series of benchmark topology optimization problems from the literature.
Highlights
As the size and complexity of numerical models used in structural analysis and design are continuously augmenting and there is no analogous increase in the available computing power, the advantages gained through the exploitation of soft computing techniques are often investigated
Within the scope of further enhancing the computational efficiency of Deep Learning-assisted Topology OPtimization (DLTOP), the authors envisioned a novel approach that combines deep learning with reduced-order modeling, which led to the development of the Deep Learning-based Model Upgrading (DLMU) scheme presented in this paper
The DLTOP methodology previously developed by the authors of [1], motivated by the insufficiency of Solid Isotropic Material with Penalization (SIMP) approach, employs deep learning approaches to reduce the number of SIMP iterations, achieving a substantial acceleration of the STO problem solution procedure
Summary
As the size and complexity of numerical models used in structural analysis and design are continuously augmenting and there is no analogous increase in the available computing power, the advantages gained through the exploitation of soft computing techniques are often investigated. The focus of the present study is on the application of soft computing methods in the fields of optimal analysis and design of structures, wherein recently there is a growing interest for exploiting deep learning techniques to accelerate topology optimization procedures. Some of the most computationally demanding problems in structural optimization are those that involve several solutions of the equilibrium equation (Equation (1)), as factorizing the stiffness matrix is time- and resource-consuming To overcome this limitation, it is necessitated to resort to techniques that reduce the order of the stiffness matrix, minimize the necessary iterations of the search process or achieve a combination of these.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
