Le Roux, Léopold, Soroka, Anthony, Ji, Ze ORCID: https://orcid.org/0000-0002-8968-9902, Kerfiden, Pierre and Bigot, Samuel ORCID: https://orcid.org/0000-0002-0789-4727 2024. Developing rapid metal AM deformations prediction using CNN. Presented at: 17th CIRP Conference on Intelligent Computation in Manufacturing Engineering, Gulf of Naples, Italy, 12-14 July 2023. Procedia CIRP. , vol.126 pp. 567-572. 10.1016/j.procir.2024.08.241 |
Preview |
PDF
- Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (1MB) | Preview |
Abstract
The production of complex 3D components using Metal Additive Manufacturing requires several steps combining different expertise, such as product design, process planning and manufacturing. This process chain would benefit significantly from faster deformation predictions enabling quicker iteration during parts design and manufacture, such as to evaluate or validate quickly a component expected final shape, internal stresses or deformation occurring during printing. This paper presents a new data-driven approach based on deep learning that can speed up the prediction of parts’ geometrical deformations by creating a digital twin of an FEA simulation. A new CNN model founded on MeshCNN and designed for the processing of AM 3D models is described and methods for testing learning capabilities using training data generated by the FEA based software Simufact Additive are proposed.
Item Type: | Conference or Workshop Item (Paper) |
---|---|
Date Type: | Published Online |
Status: | Published |
Schools: | Engineering |
Date of First Compliant Deposit: | 6 September 2024 |
Last Modified: | 26 Nov 2024 15:16 |
URI: | https://orca.cardiff.ac.uk/id/eprint/171907 |
Actions (repository staff only)
Edit Item |