About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Aluminum Cast Shop Technology Symposium and Supplier Forum
|
| Presentation Title |
Toward the Use of Furnace Cameras and Vision Transformers in the Aluminum Melting Process |
| Author(s) |
Jorge Fernandez, Sebastian Langer, Stefan Günther |
| On-Site Speaker (Planned) |
Sebastian Langer |
| Abstract Scope |
Many tasks in aluminum cast-houses are still performed near melting furnaces, where workers face high temperatures, molten metal, and other hazards. There is a strong need to reduce manual work in furnace danger zones and shift personnel from intervention toward supervision.
In this work, we demonstrate how in-furnace cameras and neural networks can be integrated into a cast-house as a step toward autonomous aluminum melting. A complete cast-house operation is abstracted as a process flowchart, and individual process steps are evaluated for their automation potential.
Particular emphasis is placed on neural networks for interpreting furnace camera images. Such systems can support recognition of process states and operator-relevant events during melting. Based on the process analysis, we identify components that can already be automated, steps that may be automated soon, and operations likely to remain manual or require supervision, providing a practical roadmap for safer, more autonomous melting operations. |
| Proceedings Inclusion? |
Planned: Light Metals |
| Keywords |
Machine Learning, Aluminum, Process Technology |