| About this Abstract |
| Meeting |
MS&T23: Materials Science & Technology
|
| Symposium
|
Additive Manufacturing Modeling, Simulation, and Machine Learning: Microstructure, Mechanics, and Process
|
| Presentation Title |
Use of Machine Learning to Identify Process-Structure-Property Relationships in PBF-LB AlSi10Mg |
| Author(s) |
Qixiang Luo, Allison Michelle Beese |
| On-Site Speaker (Planned) |
Qixiang Luo |
| Abstract Scope |
In this study, the multivariate relationships among processing conditions, microstructure, and mechanical properties of AlSi10Mg fabricated by laser powder bed fusion were identified with a combination of experimental investigation and machine learning. Experimentally, a wide range of processing parameter combinations were designed to probe a range of microstructures and defect structures in the PBF-LB AlSi10Mg. The pore structures were assessed using X-ray computed tomography, grain/sub-grain characteristics using SEM/EBSD, and mechanical properties using uniaxial tension and Vickers microhardness measurements. Machine learning was used to define the process-structure-property (PSP) relationships, by predicting each PSP feature-feature link, alongside feature importance analysis that quantified the importance of each PSP feature to each other PSP feature. This data-driven framework provides quantitative information on the complex multivariate PSP relationships, providing insight to the ICME community by suggesting the key feature that must be modeled or captured for the prediction of desired material microstructures or properties. |