Magnitogorsk Iron and Steel Works (MMK) has begun operating a machine learning–based digital system that helps workers control the oxygen levels supplied to its blast furnaces. The system recommends the optimal oxygen levels for the hot blast air in order to improve blast furnace efficiency during the production of cast iron.

Until recently, the blast furnace operator made decisions based solely on readings of discrete data. In a dynamic environment, however, those readings may not always be accurate, which could lead to erroneous recommendations. Now the recommendations are provided by an artificial-intelligence system trained on historical data.

“Experts from MMK-Informservice have developed a dynamic mathematical model for us that analyses, in an online environment, the gas-dynamics parameters and the results of laboratory measurements of the oxygen content. The system has an accuracy of 90%, enabling our specialists to monitor the iron smelting process more effectively and enhance overall process performance,” said Alexander Pavlov, the head of MMK’s blast furnace shop.

The data generated by the mathematical model is displayed on the control panels of blast furnaces Nos. 6, 9 and 10, as well as within the blast furnace recommendation systems and at the automated workstations used by blast furnace operators. The digital solution is used to correct sensor errors, improve data accuracy and ultimately lead to better decision-making during the iron smelting process.