摘要
建立了两种转炉炼钢终点氧预报模型。即数学统计模型和神经网络模型。数学模型采用多元线性回归方法建模,该模型简单、可视,但预报效果不理想,预报误差小于80×10-6,命中率仅为72.7%。神经网络模型在选取适当输入参数的基础上,通过对现场生产数据进行训练,求得合理优化的网络权重,可对转炉终点氧含量进行离线预报,该模型的预报结果较好,预报误差小于80×10-6时,预报命中率超过86.4%。
Mathematical statistics model and neural network model are described. The Mathematical statistics model is established with multiple linear regression equation. This model is simple and visible, but its prediction result is not ideal. The shoot ratio is about 72.7% when the prodietion error was less than 80×10^-6.On the basis of selecting some suitable input parameters, the network model trains the practical data to obtain the logical optimum net-weights, and then predicts the BOF end oxygen content offline. The neural network model＇s prediction result is satisfactory. The shoot ratio is above 86.4% when the prediction error was less than 80×10^-6.
出处
《山东冶金》
CAS
2006年第1期40-42,共3页
Shandong Metallurgy
关键词
转炉
终点氧预报
回归分析
神经网络
basic oxygen furnace
end point oxygen content prediction
regression analysis
neural network
