Yakovlevsky GOK Severstal reduces downtime with computer vision

2025

Monitoring the condition of ore passes

The computer vision model monitors the condition of the ore passes of Yakovlevsky GOK (part of Severstal) and helps to reduce downtime when transporting ore from the face. The solution was developed by Severstal Digital together with experts from the Yakovlevsky GOK and the Severstal-Infokom Metallurgical Development Center. Severstal announced this on September 8, 2025.

Ore passes are vertical workings along which ore is lowered from the production horizon to the recovery horizon. According to them, the raw material is first shipped to gondola cars to go to the bunkers, and then from there it rises to the surface. A computer vision model analyzes images from CCTV cameras for oversized objects on ore track lattices or overflowing them. Previously, this was controlled by the dispatcher, but due to its large load, problems in ore mines were detected with a delay, the reaction time to the incident increased, as a result, downtime was recorded.

Now, when blockages are detected and oversized objects hit the ore passes, the model displays a photo with a notification in the universal user interface "Steel Gaze." Next, the dispatcher decides, for example, to send the equipment for cleaning. As a result, thanks to the use of the model in the first half of 2025, 19.3 hours of downtime due to overcrowded ore passes were prevented, and the economic effect amounted to 2.9 million rubles.

The digital portfolio of Yakovlevsky GOK has been replenished with another solution for optimizing production processes. Now the manager can switch to more important processes, and delegate the monitoring of ore passes to the digital assistant. When creating this product, we faced some limitations in the conditions of the mine: the lack of lighting on ore mines, camera shifts limited by visibility, so we developed compensating algorithms in response. Also, when developing some solutions, we used for the first time, and they will be included in the standard functionality of our software products, "commented Svetlana Potapova, Director of Severstal Digital (Artificial Intelligence cluster).

Testing of occupational health and safety solutions

Yakovlevsky Mining and Processing Plant tests several digital solutions in the field of labor protection, industrial safety and ecology. Among the key: remote control of the combine, head lamps with video recording and every shift control over the quality of the fastening. Severstal announced this on May 16, 2025.

The mining company is gradually implementing an investment program in the field of labor protection and industrial safety. So, back in 2024, a project was launched at the Yakovlevsky GOK mine to introduce a machine vision system to identify the employee's presence under the cargo when he is transported in material storage chambers. In the first months, the project worked in a semi-manual mode of adjustment, the model was trained by specialists from the Department of Labor Safety, Industrial Safety of the Plant and colleagues from Severstal Digital to identify "false" system positives. At the same time, project statistics show that since the launch of the recognition system for the presence of workers under cargo using AI, the number of potentially fatal incidents recorded has decreased significantly. If in May 2024, 240 cases of the system were recorded, then in April of this year no more than 50 such cases.

Yakovlevsky GOK is also one of the sites for the implementation of the AI platform in the field of labor protection and industrial safety of SOVA. In 2025, it is planned to connect more than 150 cameras to it to automatically detect safety violations during routine operations by employees. In total, now more than 300 cameras have been installed in the underground mine of the Yakovlevsky GOK within the framework of the Digital Mine project.

In addition, for the first time, a model of a remote control of a combine with video surveillance of blind spots from Russian manufacturers is being tested at the Yakovlevsky GOK mine. The package includes a control unit, the remote control itself with a 40-meter cable and two cameras for better viewing. This will allow the employee to perform ore mining work at a safe distance. As of May 2025, equipment is being adjusted and connected to the network. The first combine of the kopey plant, equipped with such a system, will begin work in June.

In 2025, we will allocate more than 68 million rubles for industrial safety projects alone. To improve the safety and efficiency of operations, we launch a project for every-shift photo and video recording of the quality of fastening mine workings in accordance with work organization projects. We buy equipment for all mining sites, - said Denis Golubnichy, General Director of Yakovlevsky GOK.

In addition, a project is being implemented to provide mine employees with head lamps with a video recording system and training in machine vision to prevent dangerous actions. As of May 2025, 10 lamps are in a pilot industrial test, the developers are adjusting the system. In the target state, providing all employees engaged in underground work with such improved equipment. In addition, more than 30 loading and delivery machines of the underground fleet of the plant are equipped with cameras to ensure the control of operations by personnel.

2024: Implementation of the Alert and Visualization System

Severstal's IT team, together with Severstal's occupational health and safety function, is implementing an SOVA alert and visualization system based on computer vision technology to reduce and prevent the risks of injuries and accidents among employees. Severstal announced this on November 20, 2024.

The first stage of the project started with the introduction of ten models of computer vision on 660 cameras at the Cherepovets Metallurgical Plant (Cherepovets Steel Mill), Severstal-Metiza and Yakovlevsky GOK. Read more here.