熔敷金属力学性能人工神经网络模型的研究  

Artificial Neural Network Model for Mechanical Properties of Deposited Metals

作　　者：薛小怀[1] 国旭明[1] 钱百年[1] 于少飞[1] 杨柯[1] 訾炳涛[2] 

机构地区：[1]中国科学院金属研究所,辽宁沈阳110015 [2]清华大学机械工程系,北京100084

出　　处：《机械工程材料》2001年第11期5-7,10,共4页Materials For Mechanical Engineering

基　　金：国家 973基金资助项目 (G19980 6 15 11)

摘　　要：在试验的基础上 ,采用人工神经网络方法建立了基于BP算法的熔敷金属力学性能的预测模型 ,该模型训练结果与试验值之间有很好的对应关系 ,说明该模型能准确反映合金元素与熔敷金属力学性能之间复杂的非线性关系。用该模型研究了合金元素对熔敷金属低温韧性的影响 。A mechanical property prediction model for deposited metals is built upon the experimental data with the aid of artificial neural network (ANN) based on the BP algorithm. There are good correlations between the learining results and the experimental data. It is shown that this model is able to represent the non linear relations between the alloying elements and the mechanical properties of deposited metal accurately. Effect of alloying elements on low temperature toughness was studied by this prediction model, the analyses results were consistent with the experiments.

关 键 词：人工神经网络 熔敷金属 合金元素 力学性能 BP算法 低温韧性 

分 类 号：TG407[金属学及工艺—焊接] TP183[自动化与计算机技术—控制理论与控制工程]