AP08856141 — Development of a topological optimization method based on Deep Learning and GPU-accelerated computations for creating aerodynamic structures

Objective of the project: Development of a new TO method that will include accelerated GPU and Deep Learning methods to develop 3D printed structures optimized for multiphysics applications, in particular aerodynamic interaction applications with rigid bodies. The new method will allow designers to iterate faster with fewer errors and explore the development space more deeply.

Relevance: Recent advances in additive manufacturing are now opening up new design opportunities in many industries, including aerospace and robotics, where multi-component designs developed using old-fashioned manufacturing processes can be easily replaced with lightweight one-component designs using cost-effective single-stage 3D printing technology.

Scientific adviser: Doctor of technical sciences, Professor, Akhmetov Bakhytzhan

Results obtained: Тhe project conducted a comprehensive study on accelerating computations in topology optimization (TO). CPU-, GPU-, and deep learning-based approaches were analyzed, identifying their strengths and limitations. Deep learning models (U-Net, Res-U-Net, CNN) were developed and tested on multiple datasets, demonstrating high accuracy and significant computational speedup. Compared to the traditional SIMP method, the computation time was reduced to the millisecond level. Additionally, an aerodynamic topology optimization (ASTO) approach was developed and validated for drones and soft robotic grippers. The results confirmed the high efficiency and robustness of the proposed models. The potential for commercialization and implementation as software solutions was also explored.

List of publications with links to them

1. Maksum Y., Amirli A., Amangeldi A., Romagnoli A., Ding Y., Rustamov S., Akhmetov B. Computational Acceleration of Topology Optimization Using Parallel Computing and Machine Learning Methods – Analysis of Research Trends // Journal of Industrial Information Integration. – 2022. – Article 100352. – DOI: https://www.sciencedirect.com/science/article/abs/pii/S2452414X22000231
2. Rasulzada J., Rustamov S., Akhmetov B., Maksum Y., Nogaibayeva M. Computational Acceleration of Topology Optimization Using Deep Learning // Symmetry. – 2022. – (in print).
3. Ногайбаева М.О., Ахметов Б., Расулзаде Дж.Дж., Максум Е.А., Рустамов С. Ускорение вычислительного процесса топологической оптимизации на основе сверточной нейронной сети U-Net // Известия НАН РК. Серия информатики. – 2022. – № 3. – С. 198–213. – DOI: https://doi.org/10.32014_2518-1726_2022_343_3_198-213
4. Rasulzada J., Maksum Y., Nogaibayeva M., Rustamov S., Akhmetov B. Reduction of Material Usage in 3D Printable Structures Using Topology Optimization Accelerated with U-Net Convolutional Neural Network // Eurasian Chemico-Technological Journal. – 2022. – № 4. – DOI: https://doi.org/10.18321/ectj1471 
5. Ахметов Б. Архитектура U-Net для задач топологической оптимизации: свидетельство об авторском праве № 27543 от 28.06.2022.
6. Ахметов Б. Архитектура Res-U-Net для ускорения решения задач топологической оптимизации: свидетельство об авторском праве № 28762 от 14.09.2022.