人工神经网络在超塑性固相焊接头质量预测中的应用  

Application of Artificial Neural Network in Predicting of Superplastic Solid-Phase Welding Joint Quality

作　　者：刘珊中[1] 张柯柯[2] 杨蕴林[2] 王长生[2] 

机构地区：[1]洛阳工学院电气工程系,河南洛阳471003 [2]洛阳工学院材料科学与工程系,河南洛阳471003

出　　处：《洛阳工学院学报》2002年第1期12-14,共3页Journal of Luoyang Institute of Technology

基　　金：河南省自然科学基金资助项目 (9840 40 90 0 )

摘　　要：以 4 0Cr/T10A恒温超塑性固相焊接工艺试验数据为样本 ,经正则化预处理 ,采用径向基函数网络 ,建立由恒温超塑性固相焊接焊接工艺参数预测其接头质量的人工神经网络模型。试验结果表明 ,训练出的网络模型具有收敛快、精度高等优点 ,其完全可以应用于恒温超塑性固相焊接接头质量控制设计中。Taking the processing examination datum of isothermal superplastic solid phase welding as sample,through pretreatment of the datum,the artificial neural network(ANN) model is established with radial basis function,which is used to predict the joint quality of superplastic solid phase welding with its processing parameters.The experimental results indicate:the trained ANN model has such virtues as quick convergent speed and high precision,and can be completely used into the quality control design for superplastic solid phase welding joint.

关 键 词：人工神经网络 超塑性固相焊接 接头质量预测 应用 模型 径向基函数 正则化 

分 类 号：TG456.9[金属学及工艺—焊接]