高炉铁水含硅量神经网络预测模型

A study on neural network prediction model of Si content in hot metal

按现代控制理论,将高炉视作多输入-单输出系统。引入人工神经网络(ANN)方法,选定若干参数作为硅含量的相关变量,建立标准的三层BP网络铁水硅预报模型。用该模型对津西5#高炉的生产数据进行离线预报,允许误差为±0.1％时命中率达到81％。

Blast furnace is regarded as a system of multi-input and single-output based on modern control theory. Artificial Neural Network has been used,several variables have been selected, and a standard three layers BP( Background Propagation) network model of silicon content pridiction is set up. With production data of No. 5 BF in JinXi Iron and Steel Co. in 2000, the off-line prediction results show that the model aims at 81% with the error of ?. 1%.