基于 K⁃Means、XGBoost和PSO的高炉布料矩阵优化研究

打开文本图片集

关键词： 高炉； 布料矩阵； K⁃Means； XGBoost； 粒子群算法； 节能降中图分类号：TN86⁃34；TP18；TF54 文献标识码： A文章编号：1004⁃373X（2025）12⁃0120⁃09

Abstract：The optimization of the burden distribution matrix is a key step in realizing energy conservation and carbon reduction of blast furnaces. However， existing researches have not fully revealed the mapping relationship between the burden distribution matrix and fuel consumption parameters. On this basis， a method of burden distribution matrix optimization based on K ⁃Means clustering， extreme gradient boosting （XGBoost） and particle swarm optimization （PSO） argorithm is proposed. The K ⁃ Means and fuzzy C ⁃ means clustering algorithms are compared and analyzd in the blast furnace burden distribution matrix clustering， and the K⁃Means model with better clustering effect is selected for the cluster analysis of the blast furnace condition. The key characteristic parameters of burden distribution matrix were extracted by combining K ⁃ Means clustering results and feature selection， XGBoost， radial basis neural network and random forest model were used to predict the fuel ratio of blast furnace， and the XGBoost model with the most accurate fuel ratio prediction was selected as the prediction model. On the basis of XGBoost model， PSO and genetic algorithm were used to optimize the minimum fuel ratio and compared it， and the PSO with better optimization effect was selected for the result analysis. The results show that the proposed method can improve the melting conditions of the blast furnace ore to a certain extent， reduce the fuel ratio， and promote the energy saving and carbon reduction of the blast furnace.

Keywords：blast furnace; burden distribution matrix; K ⁃ Means; XGBoost; particle swarm optimization algorithm; energy saving and carbon reduction

0 引 言

钢铁生产全流程总能耗的 50% 以上[1]， CO2 排放量占钢铁生产碳排放总量的 70% 以上[2]。（剩余10645字）