| About this Abstract |
| Meeting |
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
| Symposium
|
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
| Presentation Title |
Physics-based Machine Learning for In-Situ Infrared Detection of Micropores in LPBF |
| Author(s) |
Berkay Bostan, Shawn Hinnebusch, David Anderson, Albert To |
| On-Site Speaker (Planned) |
Berkay Bostan |
| Abstract Scope |
The increasing benefits of machine learning (ML) are transforming the understanding, control, and mitigation of defects in manufacturing processes, particularly in Laser Powder Bed Fusion (LPBF). This study introduces an innovative ML framework that leverages physics-based in-situ infrared camera data to detect microscopic porosity in LPBF parts with high accuracy. Unlike previous approaches that focused on larger pores using simplified settings, our method excels in identifying smaller pores. When evaluated on previously unseen parts, the framework demonstrated over 90% accuracy while maintaining a false positive rate below 3%, effectively detecting pores more than ten times smaller than the sensor resolution. SHAP (SHapley Additive exPlanations) analysis was utilized to explore complex pore formation mechanisms under various conditions. This research demonstrates the crucial role of machine learning in advancing in-situ porosity detection and enhances our understanding of pore formation in LPBF processes. |
| Proceedings Inclusion? |
Definite: Post-meeting proceedings |