How can mixed metal scrap be analysed reliably and in real time? This is the question addressed by the AlloySort research project. At the beginning of December, the project participants met at AiNT in Stolberg to discuss the current status of the project and coordinate the next steps. Prof. Dr. Markus Lange-Hegermann and his research associate Helmand Shayan from inIT took part in the meeting.
The aim of AlloySort
So far, the copper and aluminium industries have lacked a method that can determine the composition of heterogeneous recycled materials in a non-destructive manner. AlloySort is developing a new approach to address this issue: prompt gamma neutron activation analysis (PGNAA) combined with AI-based evaluation methods, which will later be integrated into a conveyor belt system. This will enable sorting by both material class and alloy content.
Find out more about the project: https://www.init-owl.de/forschung/projekte/detail/alloysort-echtzeit-pgnaa-analyse-metallischer-legierungen-fuer-eine-nachgeschaltete-zielgerichtete-sortierung/
Results and Status Quo
At the project meeting, AiNT presented the current status of the measuring system and the ongoing work packages. While the mechanical construction is progressing well, the partners are currently focusing primarily on data analysis and developing the methodology.
inIT presented new results on the classification and regression of PGNAA data. Helmand Shayan demonstrated that highly noisy, short-term measurements of less than a quarter of a second can be modelled mathematically in such a way as to enable highly reliable material classification. Based on this, methods are being developed that can identify the type of alloy and estimate mixing ratios. The project partners then discussed joint optimisation approaches.
Project progress and next steps
With the results presented and the close exchange between the project partners, AlloySort is making noticeable progress toward an industry-ready inline analysis. The coming months will be dominated by plant construction, expansion of the data bases, and further validation of the mathematical models. Prof. Dr. Markus Lange-Hegermann draws a positive conclusion:
“The results so far show that PGNAA combined with data-driven methods can provide a reliable basis for analyzing alloy compositions in very short measurement times. This brings us a significant step closer to real-time capability.”