LF精炼钢水温度预测模型的开发与应用  

Development and Application of Prediction Model for Molten Steel Temperature in LF Refining

作　　者：彭翼军 尹冬航 周鹏 严笋 彭其春[2,3] 杨建[4] PENG Yijun;YIN Donghang;ZHOU Peng;YAN Sun;PENG Qichun;YANG Jian(Hunan Hualian Yunchuang Information Technology Co.,Ltd.,Changsha,Hunan 410000,China;The State Key Laboratory of Refractories and Metallurgy,Wuhan University of Science and Technology,Wuhan,Hubei 430081,China;Key Laboratory for Ferrous Metallurgy and Resources Utilization of Ministry of Education,Wuhan University of Science and Technology,Wuhan,Hubei 430081,China;Hunan Valin Xiangtan Iron&Steel Co.,Ltd.,Xiangtan,Hunan 411100,China)

机构地区：[1]湖南华联云创信息科技有限公司,湖南长沙410000 [2]武汉科技大学省部共建耐火材料与冶金国家重点实验室,湖北武汉430081 [3]武汉科技大学钢铁冶金及资源利用省部共建教育部重点实验室,湖北武汉430081 [4]湖南华菱湘潭钢铁有限公司,湖南湘潭411100

出　　处：《自动化应用》2024年第1期224-225,228,共3页Automation Application

摘　　要：针对LF工序终点钢水温度预测的问题,基于XGBoost算法建立了LF精炼钢水温度预测模型。在温度预测模型的离线测试阶段,模型在温度误差±5℃、±10℃时的命中率分别为79.8%、98.5%;模型上线运行一段时间后,统计了某段内的180炉数据,得到模型在温度误差±5℃、±10℃时的命中率分别为86.6%、97.8%。该温度预测模型满足钢厂的实际使用需求。A temperature prediction model for LF refined steel was established based on the XGBoost algorithm to address the issue of predicting the endpoint steel temperature in the LF process.In the offline testing stage of the temperature prediction model,the hit rates of the model were 79.8%and 98.5%respectively when the temperature error was±5℃and±10℃.After running the model online for a period of time,180 furnace data within a certain period were collected,and the hit rates of the model were 86.6%and 97.8%respectively when the temperature error was±5℃and±10°℃.Thistemperatureprediction model meets the actual usage needs of steel mills.

关 键 词：钢包炉精炼 温度预测 XGBoost算法 

分 类 号：TF769.2[冶金工程—钢铁冶金]