一种方形盒坯料快速设计方法
A Design Method of Square Box Blank
-
摘要: 针对冲压件坯料设计存在的难题,本文以方形盒冲压件为对象,提出一种结合径向基函数神经网络(RBFNN)和有限元反向法(LA)的快速坯料设计方法.利用神经网络高度的非线性映射能力,建立方形盒零件几何形状和坯料轮廓尺寸的映射模型.利用有限元反向法为RBFNN提供训练样本,同时采用基于变形路径的坯料优化方法改进了训练样本的精度,提高了RBFNN模型对的坯料外形的预测能力.实验证明采用RBFNN-IA方法可以实现方形盒坯料的快速设计,而且可以提高坯料的预测精度.Abstract: This paper presents a quick design method of square box blank based on RBFNN IA.The mapping model between square box geometry with blank profile is established by using neural network of highly nonlinear mapping ability.The training samples of neural network are provided by IA and in order to improve the predicting result of blank profile,the samples are optimized based on deforming path iterative method.The efficiency and precision of RBFNN-IA are proved by experiment.