基于神经网络的2D70铝合金本构关系模型的建立  

Building the Constitutive Relationship Model of 2D70 Aluminum Alloy Based on Neural Network

作　　者：周细林[1] 

机构地区：[1]江西蓝天学院材料系

出　　处：《江西蓝天学院学报》2011年第3期33-36,共4页Journal of Jiangxi Blue Sky University

摘　　要：利用Thermecmastor--Z型热加工模拟实验机对2D70铝合金进行等温恒应变速率压缩实验，获得在不同变形温度、应变速率和变形程度下的流动应力数据。用实验所得数据建立具有BP算法的三层前馈式神经网络的模型，训练结束后的神经网络即成为一个知识基的本构关系模型。误差分析表明，该神经网络本构关系具有较高的精度，可用于指导2D70铝舍金热加工工艺的制定，也可以为其它合金本构关系的建立提供参考。Isothermal and constant strain rate compression tests on 2D70 aluminum alloy are conducted by using the Thermecamastor Zsimulator to obtain the flow stress data at different temperatures, different strain rates and different deformation degrees. Based on the experimental data, an artificial neural network model with back propagation algorithm is established and the neural network after the training hence becomes the knowledge--based constitutive relations model. Error analysis shows that the artificial neural network model for constitutive relationship has a higher predicted precision, and it can be used for guiding the hot working of 2D70 aluminum alloy and providing a reference for the establishment of the constitutive relationship of other alloys.

关 键 词：2D70铝合金 本构关系 神经网络 热加工工艺 

分 类 号：TG311[金属学及工艺—金属压力加工]