摘要
转炉内的温度极高,对终点温度和碳的含量很难及时、准确地测量,因此建立精确的温度和碳的预报模型十分重要.但转炉炼钢是一个非常复杂的很难用数学方程精确描述的高温冶金反应过程,传统的静态模型控制精度不高,命中率不很理想.为此提出了基于BP神经网络的转炉炼钢终点温度及碳含量的预报模型,以Levenberg-Marquardt(LM)算法来训练网络,其算法是梯度法与高斯牛顿法的结合.仿真结果表明,预报精度高于传统的机理模型.
Steelmaking is a very complicated pyrometallurgical process, which is very difficult to be described accurately. Artificial neural network can be used for dealing with non-linear problems in metallurgical industry due to its ability. The predictive model of endpoint temperature and carbon content of basic-oxygen furnace （BOF） steelmaking based on BP neural network was put forward, and a Levenberg-Marquardt （LM） algorithm was used to train the neural network. The applied LM algorithm is a combination of the gradient decent algorithm with the Gauss-Newton algorithm. The simulated results show that the precision is higher than which based on the traditional method.
出处
《沈阳工业大学学报》
EI
CAS
2007年第6期707-710,共4页
Journal of Shenyang University of Technology
基金
辽宁省教育厅基金资助项目(202063296)
