Improving data-efficiency of deep generative model for fast design synthesis

The convolutional neural network-based deep generative model (DGM) is a powerful tool for handling image datasets that opens up strategies for fast synthesis of optimum designs at unseen boundary conditions. Existing DGMs for design synthesis are typically based on O (10000) training data, which lim...

Full description

Saved in:
Bibliographic Details
Published inJournal of mechanical science and technology Vol. 38; no. 4; pp. 1957 - 1970
Main Authors Zhang, Yiming, Jia, Chen, Zhang, Hongyi, Fang, Naiyu, Zhang, Shuyou, Kim, Nam-Ho
Format Journal Article
LanguageEnglish
Published Seoul Korean Society of Mechanical Engineers 01.04.2024
Springer Nature B.V
대한기계학회
Subjects
Artificial neural networks
Boundary conditions
Columnar structure
Conduction heating
Conductive heat transfer
Constraint modelling
Control
Dynamical Systems
Encoders-Decoders
Engineering
Industrial and Production Engineering
Machine tools
Mechanical Engineering
Neural networks
Original Article
Proper Orthogonal Decomposition
Synthesis
Topology optimization
Vibration
기계공학
Topology optimization
Data-efficient deep learning
Image processing
Reduced-order modeling
Design optimization
Online AccessGet full text

Cover

More Information
Summary:The convolutional neural network-based deep generative model (DGM) is a powerful tool for handling image datasets that opens up strategies for fast synthesis of optimum designs at unseen boundary conditions. Existing DGMs for design synthesis are typically based on O (10000) training data, which limits the engineering applications. This paper explores the feasibility of improving DGM data efficiency with O (100) training data through prior constraints. A two-stage data-efficient deep generative model (DE-DGM) is proposed which leverages the first-stage design synthesis from probabilistic proper orthogonal decomposition and the second-stage enhancement from encoder-decoder convolutional neural network. Four topology optimization cases have been adopted, including compliance minimization, heat conduction, airplane bearing bracket design, and three-dimensional machine tool column structure design. The proposed DE-DGMs could be trained with 100–200 data and synthesize the main features of the design at unseen boundary conditions. The overall computation cost of warm-start topology optimization leveraging DE-DGM predictions reduces to 36 %–58 % of the standard cases.
ISSN:1738-494X
1976-3824
DOI:10.1007/s12206-024-0328-1