About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Cast Shop Technology
|
Presentation Title |
Automated Image Analysis of Metallurgical Grade Samples Reinforced with Machine Learning |
Author(s) |
Anish K. Nayak, Hannes Zedel, Shahid Akhtar, Robert Fritzsch, Ragnhild E. Aune |
On-Site Speaker (Planned) |
Anish K. Nayak |
Abstract Scope |
Controlling metal cleanliness in primary and secondary aluminium production is critical for ensuring quality of end product and process optimisation. Solidified aluminium melt samples are typically analysed using established techniques such as Porous Disc Filtration Apparatus (PoDFA). The primary bottleneck of PoDFA analyses, the current standard approach of assessing aluminium quality, is the manual analysis of filter micrographs by metallographers. In the present study, an efficient image analysis platform based on a machine learning algorithm capable of quantifying inclusions in PoDFA filter micrographs is developed and benchmarked. Machine learning models, compared to common image analysis techniques using minimal computational resources, allow for improved performance given versatile datasets. The method is intended to enable superior cost-scaling in aluminium metal cleanliness assessments. Future implementation of these procedures will expand on the quantitative differentiation of relevant inclusion types. |
Proceedings Inclusion? |
Planned: Light Metals |
Keywords |
Aluminum, Machine Learning, Characterization |