摘要
针对冶炼过程中碳含量不能直接测定的不足,采用RBF神经网络对真空感应炉的终点碳含量进行预报.在第一次预报时,初步计算出冶炼到达终点的时间和终点的碳含量;经过二次预报进行误差校正,使结果更加精确.结合现场120组数据进行学习和预报,预报命中率较高.实验结果表明,采用该方法预报碳含量可以取得良好的效果.
Considering the deficiency of measurement in the melting process, an RBF neural network method is developed to predict the end-point carbon content in the vacuum induction furnace. It can give reliable predictions of end-point time and carbon content of molten steel in the first-round prediction. The prediction accuracy can be improved by the error correction in the second-round prediction. Total 120 set of data are used for model training and validation. The results show that the proposed method is effective.
出处
《控制与决策》
EI
CSCD
北大核心
2006年第2期210-212,216,共4页
Control and Decision
基金
国家自然科学基金项目(60374003)
国家"973"计划子课题(2002CB312200)
关键词
真空感应炉
终点碳含量预报
神经网络
误差校正
Vacuum induction furnace
End-point carbon content prediction
Neural network
Error correction