基于神经网络的X65管线钢闪光焊接头性能预测  

Predication of mechanical properties for flash butt welding joints of X65 pipeline steel based on neural network

作　　者：盛安 高世一[2] 张鸿[1] 刘正林[2] 

机构地区：[1]中南大学材料科学与工程学院,湖南长沙410083 [2]广州有色金属研究院广东省现代焊接技术重点实验室,广东广州510650

出　　处：《电焊机》2014年第5期136-140,共5页Electric Welding Machine

基　　金：国家国际科技合作专项资助项目(2011DFB70050);广东省对外科技合作专项资助项目(2012B050100014);广州市国际科技交流与合作专项资助项目(2012J5100049);广东省重点实验室建设项目资助项目(2012A061400011)

摘　　要：针对X65管线钢闪光焊接头质量受多个工艺参数影响,焊后接头力学性能难以预测的问题,采用BP(误差反向传播)人工神经网络的方法建立X65管线钢闪光焊接头力学性能预测模型,用于预测焊接接头的弯曲试验结果。确定了五个接头力学性能特征量,建立了输入层节点数为5、隐含层节点数为12的预测模型,同时对归一化公式进行改进。实际结果表明,网络的预测精度较好,能够为焊接接头力学性能预测提供参考。It is difficult to predict the mechanical properties of X65 pipeline steel welding joints and the quality of joints will be influ- enced by lots of welding technical parameters. Aiming at this problem, taking the method of BP （ Back Propagation） neural network to build the predicted model for mechanical properties of X65 pipeline steel flash butt welding joints, which can be used to predict the result of bending test. It has confirmed five characteristic features and established a predicted model containing five input layer nodes and twelve hidden layer nodes, in the mean time, the normalization formula has been improved. The actual results have been shown thal the prediction accuracy of the network is well, which can be a reference for predication of the mechanical properties of welding joints.

关 键 词：管线钢闪光焊 BP神经网络 力学性能 预测 

分 类 号：TG457.6[金属学及工艺—焊接] TG456.9