模具工业2017年第43春第12期模具CAD/CAM
基于神经网络与遗传算法的方锥件渐进
成形扭曲位移优化邱小婷，赵伟'，滕培秀”，陈丽，周白杨（1.福州大学至诚学院，福建福州350002；
2.福州大学机械工程及自动化学院，福建福州
350002）
摘要：针对渐进成形过程中工具头在板料表面切向运动时对板料进行横向推挤作用，引起板料周向运动产生扭曲变形，影响零件成形精度，采用Dynafom软件，对方锥件渐选成形过程进行数值模拟，研究扭曲变形规律，结果表明，露件底部和侧壁均产生扭曲，担曲位移至现中间大，两端小的分布，采用神经网络与遗传算法相结合的方法能较好地改善零件扭曲位移，最优工艺参数为工具头半径5.06mm。层间距1.17mm
关键词：周向位移；数值模拟；工具头半径；多目标优化；神经网络
中图分类号：TG386：0242.21
文献标识码：B
D0I: 10.16787/i.cnki.10012168.dmi.2017.12.001
文章编号：1001-2168(2017)120000105
Optimization ofdistortion displacement forincremental forming
of square cone-shaped part based on neural network and
genetic algorithm
QIU Xiao-ting', ZHAO Wei', TENG Pei-xiu', CHEN Li', ZHOU Bai-yang
(1.Zhicheng College, Fuzhou University, Fuzhou, Fujian 350002, China; 2.School of
Mechanical Engineering and Automation, Fuzhou University, Fuzhou, Fujian 350002, China) Abstract: The circumferential displacement of the parts was cased by the tool moved tan-gent on the surface of the sheet material in the incremental forming and the sheet was to be pushed in the horizontal direction. So in order to improve the precision of the parts, the incremental forming of square cone-shaped part was numerical simulated by using Dy-naformsoftwareandthelawofcircumferential
deformation was studied. The results
showed that the distortion was occurred on the bottom and walls of the parts, and the dis-tortion displacement was large in the middle but small in the end. Finally the parts were improved by combining with neural network and genetic algorithm, and the optimum param-eters were the radius 5.06 mm and layer spacing 1.17 mm of the tool.
Key words: circumferential displacement; numerical simulation; tool radius; multi-objective optimization; neural network
引言 0
渐进成形是数控加工与塑性成型相结合的新
收稿日期：2017-06-21，
基金项目：国家级大学生创新创业项目(201613470002)；福建省教育厅科技项目（JAT160660）
作者简介：邱小婷(1994-），女（汉族)，福建泉州人，主要研究方向为塑性成型及模具CAD/CAE
万方数据
1
兴如工技术，是利用工具头对板料进行加工，导致板料局部变形，工具头不断运动，板料不断发生局部变形，达到零件整体变形的塑性成型。与传统冲压不同，工具头在成形过程中沿板料表面切向运动，横向挤压板料，引起板料扭曲变形，这种现象在方形和异形零件的渐进成形中较为突出。现借助Dynaform软件，对方锥件渐进成形过程进行数值