About this Abstract |
Meeting |
2021 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2021)
|
Symposium
|
Special Session
|
Presentation Title |
Layer-wise Certification for Direct Energy Deposition Processes based on Melt Pool Morphology Dynamic Analysis |
Author(s) |
Mahathir Mohammad Bappy, Chenang Liu, Linkan Bian, Wenmeng Tian |
On-Site Speaker (Planned) |
Wenmeng Tian |
Abstract Scope |
The repeatability issue is a major barrier for wider adoption of additive manufacturing (AM) technologies in mission-critical applications. The defects will significantly compromise the mechanical properties of the fabricated parts. Therefore, there is an urgent need in reliable AM process certification. The paper proposes a new layer-wise certification method by leveraging morphology dynamics of layer-wise melt pool images. Specifically, the variability in intra-layer melt pool morphologies is formulated as the optimal transport problem, and the Wasserstein metric is used to characterize the minimum amount of work necessary to transport between two consecutive melt pools. Subsequently, multiple new layer-wise features are extracted based on those Wasserstein metrics, and supervised machine learning methods can be applied for anomaly detection. The proposed method is validated using the direct energy deposition (DED) process, which demonstrates satisfactory anomaly detection performance. |
Proceedings Inclusion? |
Definite: Post-meeting proceedings |