Abstract
In situ process monitoring is a key requirement for increased industry acceptance of powder bed Additive Manufacturing. As sensing technologies increase in maturity, attention must also be given to effective data exploration techniques. These data are often high-resolution and multi-modal, with each build consisting of thousands of layers. Here, the authors propose two methods enabling users to rapidly identify layers of interest within a build. Both methods leverage results from deep learning based segmentations of in situ powder bed images. The first method is an unsupervised “reverse layer search” algorithm while the second method uses supervised machine learning.
Original language | English |
---|---|
Pages (from-to) | 35-39 |
Number of pages | 5 |
Journal | Manufacturing Letters |
Volume | 36 |
DOIs | |
State | Published - Jul 2023 |
Funding
This research was sponsored by the US Department of Energy’s Advanced Manufacturing Office. Funding support was also provided by the Transformational Challenge Reactor program which is administered by the US Department of Energy’s Office of Nuclear Energy. The builds used in this work were printed by Chase Joslin. William Halsey provided a technical review of this document. This research was sponsored by the US Department of Energy's Advanced Manufacturing Office. Funding support was also provided by the Transformational Challenge Reactor program which is administered by the US Department of Energy's Office of Nuclear Energy. The builds used in this work were printed by Chase Joslin. William Halsey provided a technical review of this document.
Keywords
- Additive manufacturing
- Data exploration
- Machine learning