AI Boosts Ore Processing Efficiency at Russian Mining Plant

A machine-learning system at the Bystrinsky mining complex is increasing output and reducing downtime through adaptive process control.

The Bystrinsky Mining and Processing Plant in Russia’s Zabaykalsky Krai has successfully implemented an optimal control system for ore grinding, fully based on simulation and machine learning. Neural networks collect and process sensor data, displaying key equipment performance metrics. Based on this analysis, the system calculates optimal settings for the entire technological process.

The results have been tangible. Productivity in the semi-autogenous grinding mills has increased by 2.64 percent—a significant boost in concentrate output. Average downtime for scheduled maintenance has dropped, reducing overall idle time. The digital model is far from static: ore characteristics can change within a single day, and equipment conditions vary as well. As a result, the model is constantly updated, adjusting the process in real time.

Russia’s experience with adaptive grinding control could have global implications for mining. Consulting firms estimate that industry-wide adoption of such technologies could raise EBITDA by 5 to 7 percent by 2028.