摘要
通过有限元仿真分析,较准确的模拟了带钢轧制过程,获取对轧机板形影响较大的参数值,并将其结果作为训练样本对神经网络进行训练,建立了较为理想的基于神经网络的板形预测模型,实现了轧制过程中的板形参数的预报。仿真结果表明该神经网络与有限元结合的板形预测模型可获得良好的预测精度,弥补了传统板形预测模型的预测精度不能满足板形在线控制要求的缺陷。
By using finite element simulation analysis, the rolling processing of strip steel has been rather truthfully simulated and some parameters that have essential influence upon mill flatness have been obtained with the results to be served as learning sample books for exercising the neutral network, Finally, an ideal flatness prediction model, based on BP neural network, is established so that the prediction of the flatness parameters in rolling operation can be accomplished, Simulation result shows that the flatness prediction model which combines BP neural network with finite element can get a good precision in the prediction and compensates the defects that the precision of traditional flatness predicting model can＇t meet the need of on - control.
出处
《重型机械》
2007年第3期5-8,共4页
Heavy Machinery
关键词
板形预测
BP神经网络
有限元
flatness prediction
BP Neural Network
finite element
