Controlled Anisotropy Generation and Variational Autoencoder Latent Space-Based Mechanical Generative Design of Periodic Lattice Structures

Date

2025-8-21

Author

Öztaşkın, Alptuğ

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

1241
views

957
downloads

Cite This

Lattice structures of different varieties are widely used in additive manufacturing to achieve lightweight and use-case-specific structures. While extensive literature exists for the generation of new truss lattices for targeted property design via the use of variational autoencoders, no such work exists for triply periodic minimal surfaces (TPMS). Moreover, a very limited literature exists on inducing anisotropy on TPMS, which leaves an unexplored research area that could result in better-performing continuous periodic structures. Through the use of $\beta$ and conditional variational autoencoders with different means of regressor integration, this thesis explores the latent design space between various TPMS lattices and proposes different solutions for exploration through interpolation as well as inverse design of the porosity and directional elastic moduli.

Subject Keywords

TPMS, Anisotropy, VAE, Inverse Design, Lattice

URI

https://hdl.handle.net/11511/115593

Collections

Graduate School of Natural and Applied Sciences, Thesis