摘要
针对板料拉伸过程中出现拉裂、起皱等缺陷,通过人工神经网络技术研究了变压边力对矩形盒件拉伸成形效果的影响。建立了有限元模型,利用仿真软件Dynaform及"固定间隙法"获取样本数据;通过建立网络模型并对其学习训练,利用训练好的网络模型展开了对板料拉伸成形过程中变压边力预测技术的研究,获取了理想的压边力控制曲线。预测结果是板料的最大减薄率为16.2%,最大增厚率为6.6%,精度符合要求。仿真结果表明,BP神经网络可以实现对板料拉深成形变压边力的预测。
For the defects of cracking and wrinkling in the process of sheet metal stretching,the influence of variable blank-holder force in the drawing forming of rectangular box was studied by means of artificial neural network technology. The finite element model was established,and the sample data was obtained by simulation software Dynaform and＂fixed gap method＂. Through the establishment of network model and its learning and training,the prediction technology of the variable blank-holder force in the process of sheet metal stretch forming was researched by the trained network model,and the ideal curve of controlling blank-holder force was obtained. The prediction results are the largest thinning rate 16. 2% and the largest thickness rate 6. 6% of the sheet metal. The accuracy requirement is met. The simulation results show that the BP neural network can realize prediction of variable blank-holder force during the deep drawing.
出处
《锻压技术》
CAS
CSCD
北大核心
2015年第11期27-31,共5页
Forging & Stamping Technology
基金
吉林省省级经济结构战略调整引导资金专项项目(20141131)
关键词
人工神经网络
矩形盒件
变压边力
拉深成形
预测
Artificial Neural Networks（ANNs）
rectangular box parts
variable blank-holder force（VBHF）
deep drawing
prediction
