Machine Learning-Based Topology Optimization in 3D Printing

Almira Askhatova, Yerlik Gabdulla, Amanali Bekbolat, Essam Shehab, Md. Hazrat Ali

2023

Closed access Metadata (CSL-JSON)
DOI 10.1109/smartnets58706.2023.10216161
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The extremely lengthy computational times, coupled with the impracticality of the Finite Element Method (FEM)-based simulations, created a strong demand to accelerate the procedures of topology optimization involved in product development. This paper proposes a solution in the form of a deep Convolutional Neural Network (CNN) designed for 3D structural topology optimization, which predicts a structurein a relatively short amount of time without employing any iterative strategy. The methodology was created to empower users to incorporate a machine-learning approach to topology optimization in their designs while enabling the non-iterative strategy and increasing the efficiency of the process. The proposed approach is anticipated to significantly reduce computational time while improving the printed component’sstructural performance and design quality. This study offers a new perspective on topology optimization in 3D printing andhighlights the potential of machine learning in advancing theefficiency and precision of this process. The outcomes of this study illustrate that the proposed model possesses the capacity to conduct topology optimization of the cantilever beam structureat a significantly accelerated pace, as compared to specialized software showing a reduction in computation time, amounting to approximately 99%. These results hold promising implications for the time-critical engineering and optimization processes.