基于BP神经网络的中厚板层流冷却终冷温度预报  

Prediction of final cooling temperature of laminar cooling of plate based on BP neural networks

作　　者：华建社[1] 赵小龙[1] 刘明华[1] 

机构地区：[1]西安建筑科技大学冶金工程学院,陕西西安710055

出　　处：《南方金属》2011年第5期8-11,共4页Southern Metals

基　　金：陕西省工业攻关项目支助(2009K07-20)

摘　　要：通过分析层流冷却过程中钢板终冷温度的各种影响因素,建立了终冷温度的BP神经网络预报模型,确定了网络的输入、输出量.运用"试凑法"对不同网络结构进行了训练,最终确立了神经网络模型结构,并用所建模型对终冷温度进行预测.结果显示:网络预测值与现场实测值比较逼近,误差基本维持在±7℃,证明所建模型是合适的,可以用于生产现场的指导.Various factors affecting the final cooling temperature in laminar cooling of steel were analyzed in the paper. A prediction model for the final cooling temperature was proposed based on the BP neural network, and the input and output values for the network were determined. The neural network model structure was established by training different network structures using the ＂trial and error＂ method, and the final cooling temperature was predicted by the model. The results showed that the network-predicted value was in well agreement with the measured one, with error maintained at ± 7 ℃, thus indicating that the model was applicable and could be used to guide the steel production.

关 键 词：中厚板 BP神经网络 层流冷却 终冷温度 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TG335.5[自动化与计算机技术—控制科学与工程]