摘要
变压边力策略是在冲压件成型中控制回弹的一种有效措施。综合运用径向基函数(RBF)神经网络和三维回弹模拟技术开展了由冲压件成形质量反求变压边力的研究,同时研究了径向基函数神经网络的动态结构设计问题,提出一种基于泛化的径向基函数神经网络的动态结构设计方法DYNSDRBF,编制了相应的计算程序。DYNSDRBF方法在变压边力反演神经网络设计中的应用结果表明,运用该方法设计的神经网络具有较好的计算精度,可有效地提高冲压件的成形质量。
The strategy of variable blank holding force is an effective measure for the springback control in sheet metal forming processes. The Radial Basis Function（RBF） network and three-dimensional springback simulating technology were employed for variable blank holding force from the forming effects. Meanwhile, the research on the dynamic structure design of RBF network was done and the dynamic structure design method, Dynamic Structure Design of Radial Basis Function network （DYNSDRBF）, based on generalization performance was developed and programmed. The application results of DYNSDRBF in the variable blank holding force identification indicate that the structure design method can show better calculation accuracy. The research results on variable blank holding force identification show that the solving method is helpful for improving the sheet forming effects.
出处
《计算机应用》
CSCD
北大核心
2008年第2期494-498,共5页
journal of Computer Applications
基金
湖南省自然科学基金资助项目(05JJ40089)
东莞市科研专项发展基金资助项目(2006D021)
关键词
径向基函数网络
动态结构设计
泛化能力
变压边力
反演
RBF network
dynamic structure design
generalization performance
variable blank holding force
inversion
