About this Abstract |
| Meeting |
2025 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2025)
|
| Symposium
|
2025 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2025)
|
| Presentation Title |
Combining Machine Learning With Physics-Based Models for Enhanced Feedforward Control of Meltpool Area in LPBF |
| Author(s) |
Nicholas Kirschbaum, Nathaniel Wood, Chang-Eun Kim, Thejaswi Tumkur, Chinedum Okwudire |
| On-Site Speaker (Planned) |
Nicholas Kirschbaum |
| Abstract Scope |
Effective in-situ process control has long been a goal in Laser Powder Bed Fusion (LPBF), with extensive research on adjusting laser power in response to measured process signatures, such as meltpool area. However, challenges persist in making models more flexible, responding preemptively to changes in the processing environment, and accommodating inputs beyond power. In prior work, we introduced a novel power-to-meltpool-area controller using a lightweight thermal model coupled with an analytical meltpool model to modulate inputs at the vector level. However, this controller performed suboptimally in regions where the simplified model could not fully capture the dynamics. This work proposes a hybrid approach, using machine learning to leverage extended temperature field information and data from prior prints to improve the physics-based model’s predictions. This strategy improves control while preserving the ability to respond preemptively and handle non-continuous inputs. Prepared by LLNL under Contract DE-AC52-07NA27344. LLNL-ABS-2005234. |
| Proceedings Inclusion? |
Planned: Post-meeting proceedings |