石钢高炉铁水含硅量神经网络预报模型  

A NEURAL NETWORK FORECASTING MODEL OF SILICON CONTENT IN LIQUID IRON IN THE BLAST FURNACE AT SHIGANG

作　　者：郭建斌[1] 郭建国[2] 

机构地区：[1]石家庄钢铁有限责任公司,河北石家庄050031 [2]河北经贸大学计算机系,河北石家庄050061

出　　处：《金属材料与冶金工程》2006年第6期31-35,共5页Metal Materials and Metallurgy Engineering

摘　　要：按照现代控制理论,利用人工神经网络方法,把高炉视为多输入—单输出系统,结合高炉生产实际建立了石钢高炉铁水含硅量神经网络预报模型。通过引入动态步长和惯性项系数提高了网络收敛速度。采用不断更新学习样本集的方法提高了铁水含硅量预报的命中率。结果表明:在允许误差为0.1%时,命中率达到了86.67%,可以为高炉操作提供指导。Based on the modern controlling theory and the artificial neural network methods,look to the blast furnace as a system of multi-input and mono-output,according to the practical circumstances of blast furnace operation,a neural network forecasting model of silicon content in liquid iron in the blast furnace at Shigang was established.The convergent rate of the network was improved by inducting the dynamic step length and the coefficient of inertia item.The continual renewal of the specimen collection was enabled to improve the hit percentage of the forecasting of silicon content in liquid iron in the blast furnace.The results indicated that,the hit percentage of forecasting was up to 86.67% while the permissive error was 0.1%.This neural network system may provide directions for the operation of the blast furnace.

关 键 词：高炉 铁水含硅量 预报 神经网络 

分 类 号：TF533.2+1[冶金工程—钢铁冶金]