Abstract
Increasing industry acceptance of powder bed metal Additive Manufacturing requires improved real-time detection and classification of anomalies. Many of these anomalies, such as recoater blade impacts, binder deposition issues, spatter generation, and some porosities, are surface-visible at each layer of the building process. In this work, the authors present a novel Convolutional Neural Network architecture for pixel-wise localization (semantic segmentation) of layer-wise powder bed imaging data. Key advantages of the algorithm include its ability to return segmentation results at the native resolution of the imaging sensor, seamlessly transfer learned knowledge between different Additive Manufacturing machines, and provide real-time performance. The algorithm is demonstrated on six different machines spanning three technologies: laser fusion, binder jetting, and electron beam fusion. Finally, the performance of the algorithm is shown to be superior to that of previous algorithms presented by the authors with respect to localization, accuracy, computation time, and generalizability.
| Original language | English |
|---|---|
| Article number | 101453 |
| Journal | Additive Manufacturing |
| Volume | 36 |
| DOIs | |
| State | Published - Dec 2020 |
Funding
This research was partly sponsored by the Transformational Challenge Reactor (TCR) program and supported by the US Department of Energy, Office of Nuclear Energy. The authors would like to thank Dr. Derek Rose and William Halsey of ORNL, and Dr. Christopher Kantzos of CMU for many fruitful discussions regarding machine learning. Collection of data from the ConceptLaser M2 machine was performed by Chase Joslin, Keith Carver, and Frederick List III of ORNL. Collection of data from the ExOne Innovents was performed by Dylan Richardson and Desarae Goldsby of ORNL. The EOS M290 data were collected by the authors with the support of the NextManufacturing Center at Carnegie Mellon University. This research was partly sponsored by the Transformational Challenge Reactor (TCR) program and supported by the US Department of Energy , Office of Nuclear Energy . The authors would like to thank Dr. Derek Rose and William Halsey of ORNL, and Dr. Christopher Kantzos of CMU for many fruitful discussions regarding machine learning. Collection of data from the ConceptLaser M2 machine was performed by Chase Joslin, Keith Carver, and Frederick List III of ORNL. Collection of data from the ExOne Innovents was performed by Dylan Richardson and Desarae Goldsby of ORNL. The EOS M290 data were collected by the authors with the support of the NextManufacturing Center at Carnegie Mellon University. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05−00OR22725 with the US Department of Energy (DOE). Research was sponsored by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office and by the Office of Nuclear Energy. Funding was also provided by the Transformational Challenge Reactor (TCR) program. It would be inappropriate for anyone currently affiliated with ORNL’s Manufacturing Demonstration Facility to review this work. The lead author also recently graduated from Carnegie Mellon University and therefor it would also be inappropriate for anyone affiliated with either CMU’s Mechanical Engineering or Materials Science and Engineering Departments in the last five years to review this work. There are no other known conflicts of interest to disclose.
| Funders | Funder number |
|---|---|
| Frederick List III | |
| US Department of Energy | |
| William Halsey of ORNL | |
| U.S. Department of Energy | |
| Advanced Manufacturing Office | |
| Office of Energy Efficiency and Renewable Energy | |
| Office of Nuclear Energy | |
| Oak Ridge National Laboratory | |
| Carnegie Mellon University |
Keywords
- Additive manufacturing
- Convolutional neural network
- In-situ anomaly detection
- Machine learning
- Semantic segmentation
