End‐Point Prediction of Converter Steelmaking Based on Main Process Data

• Published in: Steel research international Vol. 95; no. 8
• Main Authors: Kang, Yi, Zhao, Jun‐xue, Li, Bin, Ren, Meng‐meng, Cao, Geng, Yue, Shen, An, Bei‐qi
• Format: Journal Article
• Language: English
• Published: Weinheim Wiley Subscription Services, Inc 01.08.2024
• Subjects: Carbon; Carbon content; converter steelmaking; Data analysis; Data processing; Dolomite; end‐point prediction; Flow distribution; hybrid models; influencing factors; Limestone; Oxygen; Phosphorus; Phosphorus content; Prediction models; process data; Process parameters; Steel converters; Steel making
• ISSN: 1611-3683; 1869-344X
• DOI: 10.1002/srin.202400151

In this article, main process data, notably time–series data such as lance position patterns, are analyzed during converter steelmaking, and methodologies in data processing and transforming are proposed. In this study, utilizing both the transformed key time–series and primary static process data, the influence of various process parameters on the end‐point parameters of converter steelmaking is analyzed. Furthermore, it establishes predictive models for the end‐point content of carbon (C) and phosphorus (P), as well as the end‐point temperature. In the findings, it is indicated that the end‐point carbon content and temperature are primarily influenced by the oxygen flow pattern, lime addition pattern, and key smelting parameters. The end‐point phosphorus content is mainly affected by the oxygen flow pattern, limestone addition pattern, and dolomite addition pattern. Regarding the prediction of end‐point carbon and phosphorus content, and end‐point temperature, compared to seven sub‐models, the hybrid model demonstrates an average accuracy improvement of 37.88%, 25.03%, and 31.51%, respectively, and the end‐point hit rate improves by 18.77%, 19.59%, and 20.41%, respectively. In this article, a new approach is proposed to analyze dynamic variables in the basic oxygen furnace process. Using the existing time–series data of the converter, especially data on lance position, flow pattern, and slagging pattern, a hybrid model for predicting the converter end point is established. The model interprets the importance of process influencing factors.