Scholarlink

<p>The silicon (Si) content in blast furnace hot metal is strongly controlled by high-temperature reactions between slag, metal and gaseous phases, and its control is important for stable downstream steelmaking. The excess silicon increases the heat evolution during oxygen blowing in the basic oxygen furnace, causing excessive slag formation and operational instability. From a thermochemical and process efficiency viewpoint, hot metal silicon levels of around 0.4 wt% are preferred. In the present work, a data-driven framework is developed to study the effect of key operating parameters on hot metal silicon under industrial operating conditions. Regression analysis and Artificial Neural Network (ANN) modeling are adopted to capture the non-linear interaction among process variables like hot metal temperature, silicon input from fuel and iron ore, alumina content, slag basicity indices and moisture load. To evaluate the model’s performance, cross-validation is used. The best ANN architecture has one hidden layer with 5 neurons, and minimum root means square error is 0.03146. Sensitivity analysis of the trained model offers an understanding of the relative effect of individual parameters on the silicon level of hot metal. The results are translated into an operational low-silicon strategy in an industrial blast furnace and lead to a measurable reduction of hot metal silicon levels. The results indicated that ANN-based modeling serves as a moderate predictive and decision support tool for managing the silicon level in the hot metal of the blast furnace.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Analysis of Key Factors Influencing Hot Metal Silicon and its Prediction Using Artificial Neural Networks

  • 期刊论文

  • 发表日期：

    2026年6月25日

查看全文
  • Arunabh Bhattacharjee,
  • Somnath Chattopadhyaya,
  • Binay Kumar,
  • Subhashis Kundu

摘要

The silicon (Si) content in blast furnace hot metal is strongly controlled by high-temperature reactions between slag, metal and gaseous phases, and its control is important for stable downstream steelmaking. The excess silicon increases the heat evolution during oxygen blowing in the basic oxygen furnace, causing excessive slag formation and operational instability. From a thermochemical and process efficiency viewpoint, hot metal silicon levels of around 0.4 wt% are preferred. In the present work, a data-driven framework is developed to study the effect of key operating parameters on hot metal silicon under industrial operating conditions. Regression analysis and Artificial Neural Network (ANN) modeling are adopted to capture the non-linear interaction among process variables like hot metal temperature, silicon input from fuel and iron ore, alumina content, slag basicity indices and moisture load. To evaluate the model’s performance, cross-validation is used. The best ANN architecture has one hidden layer with 5 neurons, and minimum root means square error is 0.03146. Sensitivity analysis of the trained model offers an understanding of the relative effect of individual parameters on the silicon level of hot metal. The results are translated into an operational low-silicon strategy in an industrial blast furnace and lead to a measurable reduction of hot metal silicon levels. The results indicated that ANN-based modeling serves as a moderate predictive and decision support tool for managing the silicon level in the hot metal of the blast furnace.