Research on intelligent recommendation model for blast furnace blowing parameters

Abstract

With the continuous exploitation of iron ore resources, the supply of high-quality ores is increasingly constrained, while low-grade ores reduce blast furnace iron output. Improving iron production remains a key research focus, with scholars emphasising optimisation of blast air parameters. This study proposes an intelligent recommendation model for blast air parameters to enhance iron output via parameter optimisation. Using real production data from a steel plant's No. 1 blast furnace, a whale-optimised random forest model for iron output prediction is developed, achieving 95% accuracy within ±4 tons. Additionally, an enhanced particle swarm optimisation algorithm is proposed to build an intelligent blast decision-recommendation model, which optimises blast parameters for production. Simulation and field tests show the improved algorithm outperforms the traditional PSO in accuracy and stability, boosting iron output by 5.66% per furnace on average. This research provides theoretical and technical support for intelligent blast furnace operation.