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 |
Modeling Line Width and Spatial Variations in Binder Jetting Through Neural Networks, Symbolic Regression, and Density Gradient Correlation |
| Author(s) |
Tugrul Yaylali, Shu Wang, Dylan Conover, Nathan Crane |
| On-Site Speaker (Planned) |
Tugrul Yaylali |
| Abstract Scope |
Process monitoring in binder jetting is challenging because the features of interest are difficult to observe. As a result, most studies of BJ process science and process control rely on fabricating of parts that are printed and inspected by weight, dimensions, appearance, and/or density. However, this provides slow feedback and makes it difficult to understand the impact of transient effects such as humidity changes. This paper considers the examination of single line primitives using simple optical measurements and evaluates their correlation to important process parameters such as powder bed density. Deta from over 700 samples with ten process parameters. The data is analyzed using a modified neural network with Kolmogorov-style activations and a Neural ODE architecture to generate interpretable equations. These results show promising correlation with spatial density variations in the powder bed suggesting that this is a promising strategy for online monitoring of powder bed density.
|
| Proceedings Inclusion? |
Planned: Post-meeting proceedings |