Nornickel and Reksoft Implement "smart vision" to protect production

Nornickel has deployed a neural network-based video analytics system across its facilities, designed to prevent equipment breakdowns and production stoppages, the company’s press service reported.

Foreign objects—ranging from pieces of metal to fragments of wood—frequently end up in the flow of ore. Previously, these had to be spotted manually; now, the process is automated. Cameras scan the movement of raw materials in real time to detect foreign objects, while simultaneously identifying damage to conveyor belts, such as cuts, punctures, and deep scratches.

To ensure stable operation, the cameras are equipped with self-cleaning housings and connected via a wired data transmission network, which enhances the system's overall reliability. For the convenience of staff, a web interface and dashboards have been developed: workers can monitor conditions in their specific areas, while managers receive aggregated statistics across individual workshops and the entire enterprise.

The system is currently operational at the Mayak and Komsomolsky mines, as well as the Norilsk Concentrator. Its performance significantly exceeds industry standards, with a recognition accuracy of 0.96–0.98, compared to the industry benchmark of 0.9.

Looking ahead, Nornickel plans to scale this solution to its other sites. Furthermore, the technology has potential applications well beyond the mining sector, offering value to a variety of other industries. 

Alex Shishlo, Editor in Chief of the European Bureau, Rough&Polished