| About this Abstract |
| Meeting |
2023 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2023)
|
| Symposium
|
2023 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2023)
|
| Presentation Title |
Quality Prediction of AM Processes Using Volumetric CNNs with Spatialized Representations of Structure-borne Sound Sensor Data |
| Author(s) |
Jork Groenewold, Lukas Weiser, Florian Stamer, Gisela Lanza |
| On-Site Speaker (Planned) |
Jork Groenewold |
| Abstract Scope |
The low reproducibility of process quality is a challenge in additive manufacturing processes like powder bed fusion by laser beam melting (PBF-LB/M). Therefore, in-process monitoring is often used to detect process defects such as pores. One possible measurement technique are structure-borne sound sensors, which allow the laser-induced sound emissions to be monitored. Currently, the measured signals are evaluated individually for each manufactured layer. A key drawback of this approach is that the spatial arrangement of the measured data over multiple layers is not considered. This paper presents a new approach that aims to improve model performance by better exploiting the spatial context within the data. The approach consists of three steps: (1) data-preprocessing for three-dimensional assembly (2) optimization of a volumetric convolutional neural network and (3) training of models. The overall goal is to enable the models to close a quality control loop and thus enable reproducible process quality. |
| Proceedings Inclusion? |
Definite: Post-meeting proceedings |