摘要
为了预示曲面翻边成形性能,采用有限元仿真、解析计算与人工神经网络的方法对V型零件翻边成形进行了分析.通过建立有限元模型研究了工艺参数对成形性能的影响;基于全量塑性理论及膜应变假定,推导了轴对称情况的解析计算模型;以数值模拟结果作为训练样本,建立了V型翻边成形性能预测的BP神经网络模型.研究结果表明:工艺条件对翻边成形有较大影响,其中以张角的影响最为显著;解析模型计算简便,但是只适用于零件张角较小以及相对翻边高度较小的情况;有限元仿真与人工智能相结合的BP人工神经网络模型可以快速有效地预测翻边成形性.
Finite element simulation, analytical calculation and artificial neural network are applied to predict the flanging formability of a 'V' - shaped sheet metal part. A finite element model is established to study the influence of technical parameters on formability. Then an analytical model for axisymmetric case is developed based on the total strain theory and membrane assumption. Its applicability is shown through comparison with FEM results. Finally, a BP neural network for flanging of 'V' - shaped sheet metal part is established through training FEM results. It is shown from the study that technical parameters, especially the flange angle , exert obvious influence on flanging operation; analytical model is more convenient, but only applicable to the cases with small flange angle and flange height; the BP neural network model based on finite element simulation and AI technology can quickly and effectively predict the flanging formability.
出处
《材料科学与工艺》
EI
CAS
CSCD
2004年第5期497-500,共4页
Materials Science and Technology
基金
上海市科技启明星科技跟踪计划资助项目(01QMH1411)
