Russian Students Propose an AI Model for the Metals Industry
The system tracks model degradation, signals when human intervention is needed, and automates recalibration.
Graduate students in the AI in Industry master’s program at Central University, launched by T-Bank together with mining giant Norilsk Nickel, are bringing artificial intelligence into heavy industry.
At the Talnakh Concentrator, AI is already used to improve forecasts of copper, nickel, and sulfur content in ore at different stages of processing. But without regular updates and automation, the accuracy of models used for real-time production control tends to degrade over time.
That problem was addressed by graduate student Vladimir Kuvshinov, who developed an intelligent system to monitor model degradation and automate the recalibration process. The solution reduces the risks associated with human error while improving system reliability and overall performance.
From Development to Deployment
Norilsk Nickel said Kuvshinov’s project is built on a modern open-source stack, including Python, Dash, Plotly, and MS SQL, ensuring scalability and ease of maintenance.
The student designed every component of the system himself, from the database architecture to both the backend and frontend. According to the company, the models can generate annual savings of 30–60 million rubles (about $0.4–0.7 million). In total, the AI forecasts metal content at 59 points along the production chain.
The technology is now undergoing pilot industrial operation at the plant. It is already cutting costs and improving production efficiency, the company added.
