摘要
以截面形状为目字形的7003-T4铝合金型材为研究对象,以防撞横梁质量最轻和型材拉弯成形之后的截面畸变量最小为目标,以回弹量和最大减薄率为成形质量约束,以横梁摆锤碰撞时摆锤侵入位移为刚度约束,基于有限元仿真技术和NSGA-Ⅱ多目标优化算法,对型材筋的分布、厚度分布以及拉弯成形时的预拉力大小和摩擦系数进行优化设计。研究表明:在没有补拉阶段的型材拉弯成形过程中,影响型材成形后回弹量和截面畸变量的敏感因素顺序为:预拉力大小、摩擦系数;所建立的防撞横梁质量以及摆锤碰撞刚度的多项式响应面模型和回弹量、截面畸变量以及最大减薄率的神经网络模型均具有较高的精度;通过NSGA-Ⅱ多目标优化算法获得型材结构-工艺参数的Pareto最优解。
Taking 7003-T4 aluminum alloy profile with the cross-section of‘目’shape as research object,the lightest quality of the anticollision beam and the minimum cross-sectional distortion of the profile after stretch-bending forming as the target,the springback amount and the maximum thinning rate as the forming quality constraints,and the pendulum intrusion displacement when the beam pendulum collides as the stiffness constraints,the distribution of profile ribs and thickness,pretension force and friction coefficient were optimized based on the finite element simulation technique and the NSGA-II multi-objective optimization algorithm.The results show that the sequence of sensitive factors affecting the springback and the cross-sectional distortion of profile after forming are pretension force and friction coefficient in the process of stretch-bending without the drawing stage.The polynomial response surface models for the mass of the anti-collision beam and the collide stiffness of the pendulum,and the neural network models for the springback,cross-sectional distortion and the maximum thinning rate have high precision.Pareto optimal solution of profile structure-process parameters was obtained by NSGA-II multiobjective optimization algorithm.
作者
林天豪
徐从昌
林启权
李落星
向瀚林
华家辉
LIN Tian-hao;XU Cong-chang;LIN Qi-quan;LI Luo-xing;XIANG Han-lin;HUA Jia-hui(School of Mechanical Engineering,Xiangtan University,Xiangtan 411105,China;State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body,Hunan University,Changsha 410082,China;School of Mechanical and Transportation Engineering,Hunan University,Changsha 410082,China)
出处
《塑性工程学报》
CAS
CSCD
北大核心
2019年第6期67-75,共9页
Journal of Plasticity Engineering
基金
国家自然科学基金资助项目(51575467)
关键词
拉弯成形
多项式响应面
神经网络
多目标优化
PARETO最优解
stretch-bending forming
polynomial response surface
neural networks
multi-objective optimization
Pareto optimal solution
