摘要
采用Ni-Fe-C合金作为填充金属,获得了基于TIG焊的WC-30Co/45钢焊接接头.采用人工神经网络(ANN)方法,对WC-30Co/Ni-Fe-C/45钢TIG焊过程中输入参数(焊接参数和填充金属成分)和力学抗弯强度之间的关系进行预测和分析.训练数据经过数据标准化处理,送入基于反向传播的多层前馈神经网络模型训练.并采用均方误差对模型进行误差分析.并采用训练的网络对焊接参数和填充金属成分与抗弯强度之间的关系进行预测.最后通过试验对预测结果进行了误差分析.结果表明,当采用碳含量(质量分数)0.6%或0.8%;Ni/Fe比为1.9～2.7的合金作为填充金属时可以获得较高的抗弯强度;构建的基于反向传播算法的ANN模型适用于评价WC-30Co/45钢TIG焊接头的抗弯强度,优于传统方法.
Good WC-30Co/45 steel TIG(tungsten inert gas) welded joint could be obtained using Ni-Fe-C alloy as filler metal.However,Ni-Fe-C filler metal was usually developed with the 'trial-and-error' method,which wasted a lot of time and efforts.A model was developed for analysis and prediction of correlation between input parameters(welding parameter and content of filler metal) and bend strength in WC-30Co/Ni-Fe-C/45 steel TIG welding process using artificial neural network(ANN).The model was based on multiplayer back propagation neural network and trained with data sets from experiments followed by data normalization.Mean-squared-error of this model was analyzed.The bend strength was further predicted using the trained ANN model.The results showed that when joints welded with filler metals contaning(0.6wt.%) or 0.8wt.%C,and Ni/Fe ratio in the range from 1.9 to 2.7,were obtained,and higher bend strength could be reached.The ANN model could be well used to estimate the effect of parameters on bend strength of WC-30Co/Ni-Fe-C/45 steel TIG welded joints,superior to conventional techniques.
出处
《焊接学报》
EI
CAS
CSCD
北大核心
2005年第5期41-45,共5页
Transactions of The China Welding Institution
