Scholarlink

<p>The silicon content in hot metal is a critical parameter for evaluating the efficiency of blast furnace operations and product quality. Therefore, accurately predicting and controlling the silicon content in hot metal has become a key focus in optimizing the ironmaking process. Traditional methods, which rely on empirical models and operator expertise, often lack precision and adaptability in dynamic production environments. With the advances in big data and artificial intelligence, machine learning-based methods offer a more accurate and efficient approach. This paper proposes a hybrid model combining machine learning and process optimization to optimize key parameters in blast furnace operation using historical data. The model’s prediction accuracy exceeds 90%, with improved reliability through trend consistency and weight assignment. By integrating expert experience, effective control strategies were developed and implemented in an intelligent control system. Results show a reduction in coke and fuel ratios to 352&#xa0;kg/t and 512&#xa0;kg/t, respectively, with a stability rate above 80% and a utilization coefficient of 2.85 t/m<sup>3</sup>&#xa0;d. These improvements led to significant economic benefits.</p> Graphical Abstract <p></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

A New Method for Prediction and Decision-Making of Silicon Content in Hot Metal: A Hybrid Approach Based on Machine Learning and Process Optimization

  • 期刊论文

  • 发表日期：

    2026年2月18日

查看全文
  • Yujie Zhang,
  • Xiaojie Liu,
  • Ran Liu,
  • Hongwei Li,
  • Yifan Duan,
  • Xin Li,
  • Yanqin Sun,
  • Lanjie Li

摘要

The silicon content in hot metal is a critical parameter for evaluating the efficiency of blast furnace operations and product quality. Therefore, accurately predicting and controlling the silicon content in hot metal has become a key focus in optimizing the ironmaking process. Traditional methods, which rely on empirical models and operator expertise, often lack precision and adaptability in dynamic production environments. With the advances in big data and artificial intelligence, machine learning-based methods offer a more accurate and efficient approach. This paper proposes a hybrid model combining machine learning and process optimization to optimize key parameters in blast furnace operation using historical data. The model’s prediction accuracy exceeds 90%, with improved reliability through trend consistency and weight assignment. By integrating expert experience, effective control strategies were developed and implemented in an intelligent control system. Results show a reduction in coke and fuel ratios to 352 kg/t and 512 kg/t, respectively, with a stability rate above 80% and a utilization coefficient of 2.85 t/m3 d. These improvements led to significant economic benefits.

Graphical Abstract