About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing Modeling, Simulation and Artificial Intelligence
|
| Presentation Title |
New Active Learning Methodology and its Application to Modeling AM Metallic Materials |
| Author(s) |
Madyen Nouri, Oana Cazacu, Benoit Revil-Baudard |
| On-Site Speaker (Planned) |
Madyen Nouri |
| Abstract Scope |
While 3D printing technology shows benefits in terms of raw material cost, energy consumption, and ease of manufacturing complex shapes, to realize its full potential predictive models are needed. However, mechanical data are expensive to generate. As a consequence, there are few high-fidelity constitutive models that are applicable to general loadings. Performing finite-element simulations with these models is costly in terms of computational time. While machine learning techniques have proven to provide solutions, optimizing data generation becomes crucial.
In this paper, we present a new active learning methodology designed to maximize solution efficiency by reducing the amount of simulation data needed while ensuring high model performance. The advantages of this new methodology are illustrated in a case study. It is shown that there is a significant reduction in the number of simulations required for training while achieving an excellent predictive accuracy. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Machine Learning, Computational Materials Science & Engineering |