| 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 |
Multi-Layer Temperature History Prediction During DED Process with Physics-Informed Neural Network |
| Author(s) |
Bohan Peng, Janos Plocher, Ajit Panesar |
| On-Site Speaker (Planned) |
Ajit Panesar |
| Abstract Scope |
In this work, we present a framework that applies physics-informed neural networks (PINN) in obtaining the temperature history in a multi-layer DED process. The presented work is one of the first attempts in applying PINN to the multi-layer problem. It overcomes the conflict between the intrinsic discontinuous nature of the DED process and the typical shortfall of NN in working with discontinuity with some simple but efficient techniques. The result from PINN is compared against ANSYS benchmarks to demonstrate the accuracy while achieving potential time savings for large scale parts with transfer learning. The proposed framework also promises immediate availability of thermal gradient information and capability of super-resolution which opens up pathways to alternative means for thermomechanical prediction of DED process. |
| Proceedings Inclusion? |
Definite: Post-meeting proceedings |