摘要
轧前计算是热连轧的一个重要环节,由于热连轧中影响因素多、变化范围较大,普通的数学模型不能准确地反映热连轧的生产过程。RBF神经网络在预报精度和网络冗余方面占有较大优势,将它引入轧前计算中可以提高计算效率,该方法也为研究多变量复杂工程系提出了一条新思路。
Pra-rolling calculation is one of necessary technologies, In hot strip mills,the popular mathematic model cannot reflect the real process because of disturbances and variable parameters. RBF neural networks are more competitive in this field. The sample data is collected in hot mill strip and is compared with the data which is collected by the traditional model and RBF neural network model. It is obvious that application of neural networks is more accurate and redundant. This paper also brings up a new way to make a study of complicated engineering systems of much variable.
出处
《机械制造与自动化》
2009年第6期100-101,共2页
Machine Building & Automation
基金
国家自然科学基金资助项目(编号6057305)
关键词
RBF神经网络
轧前计算
热连轧
RBF networks
pre-rolling calculation
hot strip mills