基于人工神经网络的焊缝熔敷金属力学性能预测  

FORECAST THE MECHANICAL PROPERTY OF THE DEPOSITED METAL BASED ON ARTIFICIAL NEURAL NETWORK

作　　者：李连胜[1] 魏艳红[1] 杨于银[1] 

机构地区：[1]哈尔滨工业大学,150001

出　　处：《焊接》2001年第7期18-20,共3页Welding & Joining

摘　　要：基于神经网络预测技术提出了一种新思路。本文利用前向神经网络 ,使用BP算法对熔敷金属的抗拉强度、屈服强度等力学性能指标进行研究。通过对实验数据样本进行训练 ,建立熔敷金属的抗拉强度、屈服强度与焊缝熔敷金属成分之间的隐性函数 ,用此神经网络 。This paper presented a new train of thoughts based on neural networks. In this paper, by using a back-propagation (BP) algorithm, a feed-forward neural network is employed to study the tensile strength and yield strength of the deposited metal. By training the neural network with samples, the tensile strength and yield strength of the deposited metal as the function of its chemical components is established. With this neural network, the tensile strength and yield strength of the deposited metal can be forecasted.

关 键 词：神经网络 BP算法 熔敷金属 抗拉强度 屈服强度 预测 焊缝 

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]