基于神经网络的LF炉钢液温度的预测模型  

Prediction model of molten steel temperature in LF based on neural network

作　　者：张家磊 魏志君 汪亚伟 ZHANG Jialei;WEI Zhijun;WANG Yawei(Jiangsu Yonggang Group Co.Ltd.,Suzhou 215628,China)

机构地区：[1]江苏永钢集团有限公司,江苏苏州215628

出　　处：《现代交通与冶金材料》2022年第2期84-87,共4页Modern Transportation and Metallurgical Materials

摘　　要：针对现有炼钢厂在LF炉钢液温度的控制方面,大多采用人工测温的方式,存在测量精度差,调节时间长等缺点,不利于钢的质量稳定且不利于降低成本。以42CrMo4钢种为例,通过进行相关性分析,筛选出影响42CrMo4的LF炉终点温度的因素,以这些影响因素作为输入,LF炉终点温度作为输出建立了神经网络的LF炉温度预测模型,探讨给定这些输入量,是否能得到较为准确的终点温度,实验表明神经网络模型能够很好地预测LF炉钢液温度。For the existing steelmaking plants in the LF molten steel temperature control,most of them use manual temperature measurement,which has disadvantages such as poor measurement accuracy and long adjustment time. It is not benefit to the stability of steel quality and reducing costs. In this paper,42CrMo4 steel grade is taken as an example,through correlation analysis,the factors affecting the end temperature of 42CrMo4 LF are screened out. Using these influencing factors as input and the end temperature of LF as output,a neural network LF temperature prediction model is established. It is explored whether a more accurate endpoint temperature can be obtained given these input quantities.Experiments show that the neural network model can predict the temperature of molten steel in the LF very well.

关 键 词：LF炉终点温度 神经网络:人工测温 42CrMo4 温度预测 

分 类 号：TF703.5[冶金工程—钢铁冶金] TP183[自动化与计算机技术—控制理论与控制工程]