摘要
设计了用于三维拉弯成形的、可重构的柔性模具,并采用支持向量回归机和有限元模拟对柔性三维拉弯成形的回弹进行预测.使用有限元法分析了对回弹量影响较大的6个因素(包括材料参数、几何参数和工艺参数),以及它们对回弹的影响趋势.选用这6个参数设计有限元三维拉弯模拟实验,并用模拟结果训练和检验支持向量回归机回弹预测模型.通过与广泛应用的神经网络预测方法的预测值和有限元模拟试验结果的比较,检验该回弹预测模型的准确性.研究发现,该模型与神经网络相比具有更高的准确度,在试验中根据该模型预测的回弹量对模具型面进行相应的补偿,可以有效地减小回弹和形状偏差.
In this paper,first,a reconfigurable flexible die for the three-dimension stretch bending forming is designed,and the springback of profiles during the forming is predicted by means of the support vector regression and the finite element simulation. Then,six factors that greatly affect the springback magnitude（ including material parameters,geometrical parameters and process parameters） are analyzed by using the finite element method,and their impact trends on the springback are also investigated. Moreover,these six factors are employed to design a simulation of three-dimension finite-element stretch bending,and the simulated results are used to train and test the springback prediction model based on the support vector regression machine. Finally,for the purpose of verifying the proposed apringback prediction model,the predicted results are compared with those obtained by the widelyused neural network forecasting method and the finite element simulation. It is found that the proposed model is more accurate than the neural network-based method,and that,in experiments,suitable compensations to the die shape according to the springback value predicted by the model may effectively reduce the springback and the shape deviation.
出处
《华南理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2015年第2期107-113,共7页
Journal of South China University of Technology(Natural Science Edition)
基金
国家工信部重点产业振兴和改造技术专项(吉工信投资[2011]350)~~
关键词
型材
回弹预测
支持向量回归机
人工神经网络
三维拉弯成形
profile
springback prediction
support vector regression machine
artificial neural networks
three-dimension stretch bending forming
