论文部分内容阅读
针对圆筒形件拉深成形时压边力这一主要参数,通过数值模拟实验可获得大量模拟数据;在此基础上将神经网络技术及设计方法引入,并以MATLAB数学软件为二次开发平台,开发了基于BP神经网络的压边力优化模块。再将圆筒形件成形数值模拟数据作为训练样本,对训练后的网络进行测试。结果显示,网络误差小于4%,这验证了所建网络模型的正确性和实用性,为拉深成形压边力优化系统的应用奠定了基础。
Aiming at the main parameters of blank holder force during forming process of cylindrical part, a large amount of simulation data can be obtained through numerical simulation experiment. On the basis of this, the neural network technology and design method are introduced, and the matlab software as secondary development platform , Developed a BP optimization module based on BP neural network. Then, the numerical simulation data of the cylindrical part forming are used as training samples to test the trained network. The results show that the network error is less than 4%, which verifies the correctness and practicability of the proposed network model and lays the foundation for the application of the deep drawing forming BHF system.