摘要
为消除同批次金属板材参数波动对其折弯成型精度造成的影响,提出了一种基于力反馈的折弯角度实时控制方法。利用兰贝格-奥斯古德方程获得板材理想应力-应变曲线,在此基础上改变方程参数与板材厚度模拟其参数波动,生成输入样本。使用有限元分析软件ABAQUS对该样本进行仿真,获取了不同参数板材在使用相同模具时,上模受力随位移变化曲线,以及成型角度随下压行程变化曲线。建立循环神经网络模型,将上模位移与受力作为输入,成型角度作为输出,得到了根据受力变化预测成型角度的网络模型。通过仿真实验,成型角度误差为±0.6°,验证了该方法的可行性,为后续实现利用成型角度历史数据对预测模型进行智能优化的折弯控制算法奠定了基础。
In order to eliminate the influence of the parameter fluctuation of the same batch of sheet metal on its bending accuracy,a real-time bending angle control method based on force feedback was proposed.The ideal stress-strain curve of the plate was obtained by using the Ramberg-Osgood equation.On this basis,the parameters of the equation and the thickness of the plate were changed to simulate the parameter fluctuation and generate the input sample.The finite element analysis software ABAQUS was used to simulate the samples,and for the plates of different parameters,the curves of the upper die force with the displacement and the forming angle with the pressing stroke were obtained subject to the same die.A recurrent neural network model was established,which takes the displacement and force on the upper die as the input and the forming angle as the output.A network model was obtained to predict the forming angle according to the change of force.Through the simulation experiment,the error of forming angle is±0.6°,which verifies the feasibility of the proposed method,and lays a foundation for the subsequent realization of the bending control algorithm that uses the historical data of forming angle to intelligently optimize the prediction model.
作者
赵海文
鲁泽逸
张晓宇
王佳阔
ZHAO Haiwen;LU Zeyi;ZHANG Xiaoyu;WANG Jiakuo(School of Mechanical Engineering,Hebei University of Technology,Tianjin 300401,China)
出处
《现代制造工程》
CSCD
北大核心
2023年第1期104-109,115,共7页
Modern Manufacturing Engineering
关键词
V型折弯
力反馈
参数波动
循环神经网络
角度预测
V-bending
force feedback
parameter fluctuation
recurrent neural network
angle prediction
