基于遗传神经网络的拉延筋参数辨识

The Identification of Geometric Parameters of Drawbead Based on GA-RBF Neural Network

拉延筋是板料冲压成形质量的一个重要影响因素,设计和布置拉延筋是冲压模具设计的关键。研究反映拉延筋成形效果的3个参数与拉延筋几何参数之间非线性映射关系,采用遗传算法优化的反向传播神经网络构建模型,并对模型进行参数辨识。仿真实验验证了此方法的有效性,可为拉延筋的设计提供帮助。

The drawbead plays a very important role in sheet metal forming.The design of drawbead is an important part of stamping die.The application of neural network combined with genetic algorithm applied to identify the geometric parameters of drawbead in forming processes is discussed in this paper.The structure of the back propagation（BP） neural network,is optimized by genetic algorithm（GA）,is employed to construct the nonlinear mapping relation between parameters used to reflect metal forming effect and geometric parameters of drawbead.The experiment results show the validity of the proposed model which is helpful to shorten the design of drawbead.