拼焊板方盒件拉深成形过程变压边力曲线预测  

Variable blank holder force curve prediction in deep drawing for square box of tailor-welded blanks

作　　者：官英平[1] 马向东[1] 张登谦 

机构地区：[1]燕山大学先进锻压成形技术与科学教育部重点实验室,河北秦皇岛066004 [2]长城汽车股份有限公司,河北保定071000

出　　处：《塑性工程学报》2017年第2期17-21,共5页Journal of Plasticity Engineering

基　　金：国家自然科学基金资助项目(51275444);河北省自然科学基金钢铁联合研究基金资助项目(E2014203271);教育部高等学校博士学科点专项科研基金资助项目(20121333110003)

摘　　要：基于有限元数值模拟软件LS-DYNAFORM,对拼焊板方盒形件拉深成形进行模拟研究。通过改变拉深成形过程中压边力这一最重要且易于控制的工艺参数,寻求拼焊板方盒形件拉深成形时较优的变压边力曲线加载形式。为预测不同工艺参数下拼焊板方盒形件拉深成形时的较优压边力加载曲线,建立了变压边力的BP神经网络预测模型,并将该模型预测的结果与数值模拟得到的结果进行对比分析。研究结果表明,拼焊板薄板采用变压边力、厚板采用恒定压边力、且薄板压边力不小于厚板压边力的加载形式,拼焊板成形件整体质量较好,焊缝移动量较小;神经网络预测模型能较好的预测拼焊板方盒形件拉深成形时的变压边力,与数值模拟结果的最大相对误差在12.3%以内。Deep drawing of tailor-welded blanks （TWBs） for square box was numerically researched based on the finite element simulation software LS-DYNAFORM. The better variable blank holder force （BHF） loading curve was sought by changing the BHF which is the most important and easily controlled parameter in deep drawing. The BP neural network of variable BHF prediction model was established to predict the better BHF of TWBs deep drawing processes with different forming parameters, and then the comparison was conducted between the prediction results and the simulation ones. The results show that the TWBs forming part has better overall quality and less welded line movement when the thick and thin side adopt variable and constant BHF, respectively. The neural network prediction of variable BHF has a good agreement with the simulation results and the relative error is less than 12. 3%.

关 键 词：拼焊板 拉深成形 数值模拟 变压边力预测 神经网络 

分 类 号：TG386.31[金属学及工艺—金属压力加工]