基于人工神经网络的高炉渣黏度预报模型  

Development of BF Slag Viscosity Prediction Model Based on the Artificial Neural Network

作　　者：杜洪缙[1] 储滨[1] 傅元坤[2] 吴亮亮[2] 黄志甲[3] 

机构地区：[1]宝钢不锈钢有限公司炼铁厂,上海200431 [2]安徽工业大学冶金与资源学院,安徽马鞍山243000 [3]安徽工业大学建筑与工程学院,安徽马鞍山243000

出　　处：《安徽工业大学学报（自然科学版）》2013年第3期322-327,共6页Journal of Anhui University of Technology（Natural Science）

摘　　要：针对宝钢不锈钢高炉高铝渣流动性较差,且黏度预报精确度较低的现状,采用旋转型黏度测定方法,测定1 500,1 475,1 450℃,Al2O3质量分数为15%-19%,MgO质量分数为7%-11%,碱度R2为1.1-1.3条件下的高铝渣系黏度,构建炉渣黏度与其成分的数据库,在此基础上,建立人工神经网络的炉渣黏度预报模型。应用结果表明,基于人工神经网络的炉渣黏度预报模型的预报误差基本在5%以内,命中率为90%以上。According to the current situation that the mobility of high aluminum BF slag of stainless steel in Baosteel is poor, and the traditional method of slag viscosity prediction has more errors. The viscosity data of slag with high aluminum content （w（A1203）= 15%-19%, w（MgO）=7%- 11%, R2= 1.1-1.3）, was measured in experiment, at the same time, a viscosity prediction model which based on artificial neural network was developed. The applied results show that the new model prediction error is within 5% with hit rate more than 90%.

关 键 词：高铝渣 黏度 神经网络 熔化性温度 

分 类 号：TF534.2[冶金工程—钢铁冶金]