摘要
以汽车前围板为分析对象,利用DYNAFORM仿真软件建立前围板有限元模型,对其回弹进行了仿真模拟。在此基础上,在Matlab环境下建立BP神经网络回弹预测模型,对前围板在不同冲压条件下的回弹进行定量预测,并在固定板材厚度的条件下对BP神经网络进行修正,设定试验参数组进行多点回弹预测。将BP神经网络回弹预测结果与仿真测量值进行对比,结果表明,采用BP神经网络预测回弹具有较高的精度,比仿真预测耗时少,可为车身覆盖件冲压模具设计提供依据。
The simulation of the springback of the cowl panel was expounded by the establish of the cowl panel finite element model using Dynaform.The espringback prediction model of the BP neural network was established.The espringback of the cowl panel was predicted quantitatively about different stamping.The BP neural network in the condition of fixed plate thickness was corrected,and the parameters to predict the springback multiple point was set.By comparative analysising and simulation measurements,the results show that using BP neural network can predict rebound with high precision,take less time,and provide a basis for automobile body panel stamping die design.
出处
《热加工工艺》
CSCD
北大核心
2012年第9期100-103,共4页
Hot Working Technology
关键词
回弹预测
神经网络
成形仿真
模具设计
prediction of springback
neural network
forming simulation
mold design
