DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN

DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN

Bibliographic Details
Main Author: Rawat, Sharad
Language:English
Published: The Ohio State University / OhioLINK 2019
Subjects:
Mechanical Engineering
Computer Science
Topology Optimization
Deep Learning
Engineering Design
Online Access:http://rave.ohiolink.edu/etdc/view?acc_num=osu1559560543458263
id ndltd-OhioLink-oai-etd.ohiolink.edu-osu1559560543458263
record_format oai_dc
spelling ndltd-OhioLink-oai-etd.ohiolink.edu-osu15595605434582632021-08-03T07:11:21Z DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN Rawat, Sharad Mechanical Engineering Computer Science Topology Optimization Deep Learning Engineering Design Topology design optimization offers a tremendous opportunity in design and manufacturing freedoms by designing and producing a part from the ground-up without a meaningful initial design as required by conventional shape design optimization approaches. Ideally, with adequate problem statements, to formulate and solve the topology design problem using a standard topology optimization process, such as SIMP (Simplified Isotropic Material with Penalization) is possible. However, in reality, an estimated over thousands of design iterations is often required for just a few design variables, the conventional optimization approach is, in general, impractical or computationally unachievable for real-world applications significantly diluting the development of the topology optimization technology. There is, therefore, a need for a different approach that will be able to optimize the initial design topology effectively and rapidly. In this study, a novel topology optimization approach based on Generative Adversarial Networks (GAN) and Conditional Wasserstein Generative Adversarial Networks (CGAN) is developed to replicate the conventional topology optimization algorithms in an extremely computationally inexpensive way. GANs consists of a generator and a discriminator, both of which are deep convolutional neural networks (CNN). A minimax game is conducted between the generator and the discriminator as part of training where the discriminator maximizes the loss function whereas the generator tries to minimize the loss function of the model. Once trained, the generator from GAN can produce 2D/3D structures in a computationally inexpensive process instantaneously. The corresponding input variables of the new generated structures are evaluated using a trained convolutional neural network. The limited samples of data, quasi-optimal planar structures, needed for training purposes are generated using the conventional topology optimization algorithms. With CGANs, the topology optimization conditions can be set to a required value before generating samples. Moreover, CGAN truncates the global design space by introducing an equality constraint by the designer. The results are validated by generating an optimized planar structure using conventional algorithms with the same settings. A proof of concept is presented which is known to be the first such illustration of the fusion of GANs and CGANs and topology optimization. 2019-10-23 English text The Ohio State University / OhioLINK http://rave.ohiolink.edu/etdc/view?acc_num=osu1559560543458263 http://rave.ohiolink.edu/etdc/view?acc_num=osu1559560543458263 unrestricted This thesis or dissertation is protected by copyright: some rights reserved. It is licensed for use under a Creative Commons license. Specific terms and permissions are available from this document's record in the OhioLINK ETD Center.
collection NDLTD
language English
sources NDLTD
topic Mechanical Engineering
Computer Science
Topology Optimization
Deep Learning
Engineering Design
spellingShingle Mechanical Engineering
Computer Science
Topology Optimization
Deep Learning
Engineering Design
Rawat, Sharad
DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN
author Rawat, Sharad
author_facet Rawat, Sharad
author_sort Rawat, Sharad
title DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN
title_short DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN
title_full DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN
title_fullStr DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN
title_full_unstemmed DEEP LEARNING BASED FRAMEWORK FOR STRUCTURAL TOPOLOGY DESIGN
title_sort deep learning based framework for structural topology design
publisher The Ohio State University / OhioLINK
publishDate 2019
url http://rave.ohiolink.edu/etdc/view?acc_num=osu1559560543458263
work_keys_str_mv AT rawatsharad deeplearningbasedframeworkforstructuraltopologydesign
_version_ 1719455799446601728

Similar Items