基本变形弹塑性问题的人工神经网络解法探讨  

Solution on the Elastic-Plastic Problem for Simple Transformation Based on Artificial Neural Network

作　　者：何芝仙[1] 孙萍[1] 李震[1] 

机构地区：[1]安徽工程科技学院机械系,安徽芜湖241000

出　　处：《试验技术与试验机》2006年第1期24-27,共4页Test Technology and Testing Machine

基　　金：安徽省自然科学基金项目(03042306)

摘　　要：利用体积不变原理,通过拉伸试验测量真实应力应变关系的试验数据,再利用人工神经网络(BP网络)进行逼近,从而得到用BP网络表示的应力应变本构关系。并以梁的纯弯曲和圆轴的扭转弹塑性问题为例,提出了基于人工神经网络的求解新方法,计算结果验证了其方法的合理性。In this paper test data of real stress-strain relationship are acquired by tensioning test according to the law of volume invariable. A back propagation network is applied to build up a real stress-strain relationship by training the network with the test data. A novel method related to the BP network is presented and it can be solved some elastic-plastic problems It has been seen that the method is available according to the examples such as pure bending of a beam and torsion of a shaft.

关 键 词：应力应变 人工神经网络 弹塑性 塑性极限载荷 

分 类 号：O344.1[理学—固体力学]