论文部分内容阅读
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。
Aiming at the flaws such as cracking and wrinkling during the sheet drawing process, the influence of blank holder force on the stretch forming of rectangular box was studied by artificial neural network. The finite element model was established, and the sample data was obtained by Dynaform and “Fixed Gap Method”. Through the establishment of the network model and its learning and training, Edge force prediction technology research, access to the ideal blank holder force control curve. The prediction result is that the maximum thinning rate of sheet metal is 16.2% and the maximum thickening rate is 6.6%, the accuracy meets the requirements. The simulation results show that the BP neural network can predict the deformation of the blank holder during forming.