Modularization Method Based on New Layout Design in Conceptual Design Stage - Application of Multi-material Lightweight Structures Utilizing Machine Learning
Year: 2019
Editor: Harold (Mike) Stowe; Tyson R. Browning; Steven D. Eppinger; Jintin Tran; Paulo Montijo
Author: Asaga, Yasuo; Nishigaki, Hidekazu
Series: DSM
Institution: Toyota Central R&D Labs.
Section: New Methods and Algorithms
Page(s): 10
DOI number: https://doi.org/10.35199/dsm2019.5
ISBN: 978-1-912254-06-4
Abstract
By creating a frame structure of a new layout and dividing it into modules with little coupling with each other using topology optimization and clustering, we have found out the assembly units of the product and the assembly process with less rework. Replacing with a lightweight material is an effective method for reducing the weight of the structure, but in order to minimize the influence such as adhesiveness and difference in coefficient of thermal expansion due to the use of different materials, it seems reasonable to replace by module as a functional unit. Therefore, we constructed a system for structural evaluation by material replacement and the cross section that maintains equivalent stiffness for each module. We present the method of constructing the system and the effectiveness using machine learning confirmed by applying it to the box structure.
Keywords: modular design, machine learning, frame structure, topology optimization, multi-material