IEEE Access (Jan 2024)
Research on Digital Morphogenesis and Sustainability of 3D Printing Bionic Materials Based on Convolutional Neural Networks
Abstract
This study applies convolutional neural networks (CNNs) and digital morphogenesis research methods to perform biomimetic design of the morphology of 3D printed materials, furthering structural innovation based on the lightweight sustainability of biomimetic materials. Natural two-dimensional forms such as leaf veins, spider webs, and dragonfly wings are selected for digital reconstruction into three-dimensional biomimetic forms. This process involves transferring the material properties and structural advantages of natural two-dimensional biological forms to three-dimensional models. Hence, digital methods are employed to create three-dimensional representations of leaf veins, spider webs, and dragonfly wings while preserving their structural performance advantages observed in nature. CNNs style transfer technologies are utilized, employing 53 cross-sectional images of 3D models as content images for the style transfer algorithm, alongside natural two-dimensional form images as style images. This allows for the parametric reconstruction of three-dimensional biomimetic models. Finally, a series of mechanical and material performance tests are conducted to validate the mechanical and structural performance of 3D printed biomimetic structural morphologies. This study presents a research methodology for the digital reconstruction of natural two-dimensional forms into three-dimensional representations and innovatively applies digital technologies such as CNNs to material morphology research. Through the application of digital morphogenesis research methods, this study explores the sustainability and innovation of 3D printed materials.
Keywords