| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
| Presentation Title |
Generative Property Optimization of Stochastic Microstructures |
| Author(s) |
Patxi Fernandez-Zelaia, Jason Mayeur, Jiahao Cheng, Guannan Zhang, Neil Zhang, Amir Ziabari |
| On-Site Speaker (Planned) |
Patxi Fernandez-Zelaia |
| Abstract Scope |
Materials inverse design problems are essential across a range of energy applications. Experimental methods can be time consuming and physics-based modeling is generally limited to solving forward problems. Machine learning based generative models have been demonstrated to be well suited for solving inverse solutions. Denoising diffusion probabilistic models have exploded in popularity with a number of successful materials specific applications. However, there are currently no works demonstrating property guided microstructure image generation. This work establishes a generative model for two-phase composite structures with control over both feature statistics and resultant properties. The latter is achieved using an auxiliary structure-property physics surrogate network. These results show that denoising diffusion probabilistic models trained to encode structure-property relations are powerful engineering tools useful for solving inverse materials design problems. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Machine Learning, ICME |