矩形截面型材三维拉弯成形的回弹预测

发布时间：2018-05-13 08:33

本文选题：型材 + 回弹预测　；参考：《华南理工大学学报(自然科学版)》2015年02期

【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.
[Abstract]:A reconfigurable flexible die for 3D bending is designed, and the springback of flexible 3D bending is predicted by support vector regression machine and finite element simulation. Using the finite element method, six factors (including material parameters, geometric parameters and technological parameters) which have a great influence on the springback amount are analyzed, and the trend of their influence on the springback is also analyzed. These six parameters are selected to design the finite element three-dimensional tension and bending simulation experiment, and the results of simulation are used to train and test the springback prediction model of support vector regression machine. The accuracy of the springback prediction model is verified by comparing the prediction value with the widely used neural network prediction method and the finite element simulation test results. It is found that the model has higher accuracy than neural network. The springback and shape deviation can be effectively reduced by compensating the die surface according to the springback predicted by the model.
【作者单位】：大连理工大学汽车工程学院∥工业装备结构分析国家重点实验室;吉林大学材料科学与工程学院;
【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)~~
【分类号】：TG306

【参考文献】