AI Safeguards the Conveyor Line
The Komsomolsky metal mine is introducing a machine vision system designed to protect conveyor equipment from damage. The technology is currently being trained and is expected to raise the accuracy of detecting hazardous objects to more than 90%.
The Komsomolsky metal mine, part of PJSC MMC Norilsk Nickel, has introduced a system that captures images of material moving along the conveyor. An AI module analyzes the images and identifies foreign objects such as wood, large ore fragments, and similar debris. When such items are detected, the system sends a signal and the conveyor is stopped so the threat can be removed. The goal of the project is to minimize wear on costly equipment and shift operations from emergency repairs to predictive maintenance.
Testing of the machine vision system began at the end of last year. At present, the system’s effectiveness stands at 50%. The machine vision software can already distinguish between classes of foreign objects and contaminants. In the near future, its accuracy is expected to rise to more than 90%. Before the system was introduced, the entire conveyor flow was monitored by operators. As early as this year, the human factor is expected to be fully removed from the list of risks affecting the production process.
Specialists from the Diagnostics Center of the Polar Division are involved in training the AI module as supervisors and analysts. Over the past 15 years, they have compiled specialized catalogs of conveyor belt defects, which are now being used to train the AI.
A Broader Industry Trend
Although this experience is not technologically unique on a global scale, for Russia it is an important step in applying AI to heavy industrial processes.
Norilsk Nickel itself uses this technology across different operations. At the Talnakh concentrator, for example, computer vision monitors foam parameters and analyzes video streams to stabilize flotation unit performance, reduce process fluctuations, and increase metal recovery.
A digital converter assistant is also used at the Copper Plant. AI models analyze material balances, the composition of fluxing materials, and melt spectra. Operators receive recommendations on dosage and blowing time, which helps maintain product quality.
Predictive analytics are in operation at Kola MMC. The system processes thousands of parameters across the production chain in real time, detects deviations, and automatically adjusts operating modes. This makes process control more flexible and helps prevent emergency shutdowns.
An AI advisor at the Nadezhda Metallurgical Plant helps maintain the required charge composition and product characteristics while minimizing production deviations.
AI is also used to simulate drilling and blasting operations on the basis of historical and exploration data. Modeling drilling, explosive loading, detonation, and blast energy distribution makes it possible to determine well placement patterns, optimal depth and spacing, as well as the power and location of the explosives.
According to the company, AI solutions generate up to 1% in additional annual revenue and add $70 million to $100 million to operating profit.
Collective Intelligence in Mining
The experience of other market participants also clearly points to the benefits of applying AI.
The diamond mining company ALROSA uses a truck loading control system. A combination of machine vision, laser scanning, and neural network algorithms makes it possible to load mining haul trucks with rock more efficiently. As a result, the productivity of mining trucks increased by 10% at the Udachny mine and the Yubileyny open pit. In geological exploration, neural networks were trained on a digitized archive of the company’s geological data. The training set includes information collected over 50 years. As a result, a pilot project at the Alakit-Markha kimberlite field in the Mirny district of Yakutia identified promising zones for the discovery of new kimberlite bodies.
Researchers at the Seversk Technological Institute of NRNU MEPhI developed an AI-based decision support system for Dalur that plans maintenance and recovery work on process wells at uranium in-situ leaching sites. The system takes 15 parameters into account.
Vostsibugol has been implementing its Tsifrovoy razrez (Digital Open-Pit Mine) project since 2013. Among the AI solutions already deployed are systems that monitor the use of personal protective equipment by employees and track equipment condition.
Overall, AI and digitalization are becoming an important and inseparable part of mining industry modernization. The deployment of the AI module at the Komsomolsky metal mine reflects a real production transformation in which neural networks are being integrated into Russia’s heavy industry.
