Scholarlink

<p>The Ruhrstahl–Heraeus (RH) is the most important reactor to produce interstitial-free (IF) steel, and decarburization is the key step in the RH vacuum refining process. In the past several decades, a mechanistic model has been very popular to predict the decarburization process, but how to accurately determine the fitting parameters in the mechanistic model is a critical unresolved issue. Thus, a mechanistic and data-driven hybrid model is proposed to solve this issue. In the hybrid model, the mechanistic model is employed to generate datasets within parameter ranges constrained by empirical equations, and the random forest model is applied to predict the fitting parameters. Under the current operation conditions, the fitting parameters are: ak<sub>C</sub> = 0.1738 and ak<sub>O</sub> = 0.1132 in the case of natural decarburization, ak<sub>C</sub> = 0.1417, ak<sub>O</sub> = 0.0984, and oxygen absorption efficiency is 0.5688 in the case of forced decarburization. The ratios of ak<sub>O</sub>/ak<sub>C</sub> are 0.6513 in the case of natural decarburization and 0.6944 in the case of forced decarburization. They are close to the 0.69 proposed by Suzuki et al.</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

Mechanistic and Data-Driven Hybrid Model for RH Decarburization Process

  • 期刊论文

  • 发表日期：

    2026年7月27日

查看全文
  • Yuanxin Jiang,
  • Hong Lei,
  • Yili Sun,
  • Denghui Li,
  • Tianyu Zhang,
  • Meng Qu

摘要

The Ruhrstahl–Heraeus (RH) is the most important reactor to produce interstitial-free (IF) steel, and decarburization is the key step in the RH vacuum refining process. In the past several decades, a mechanistic model has been very popular to predict the decarburization process, but how to accurately determine the fitting parameters in the mechanistic model is a critical unresolved issue. Thus, a mechanistic and data-driven hybrid model is proposed to solve this issue. In the hybrid model, the mechanistic model is employed to generate datasets within parameter ranges constrained by empirical equations, and the random forest model is applied to predict the fitting parameters. Under the current operation conditions, the fitting parameters are: akC = 0.1738 and akO = 0.1132 in the case of natural decarburization, akC = 0.1417, akO = 0.0984, and oxygen absorption efficiency is 0.5688 in the case of forced decarburization. The ratios of akO/akC are 0.6513 in the case of natural decarburization and 0.6944 in the case of forced decarburization. They are close to the 0.69 proposed by Suzuki et al.