摘要
通过不同预热温度、链速、焊接温度和焊接时间下的软钎焊工艺试验,将所获得的焊点平均扩展率测试数据作为神经网络的训练样本和验证样本,构建了4×20×8×1四层神经网络,对软钎焊焊点成形进行预测,并对网络模型的预测精度进行研究。结果表明,利用该四层神经网络可对软钎焊焊点进行高精度的成形预测。
The average rates of soldering forming with different preheating temperature, chain speed, soldering temperature and soldering time were measured, and the sample data were used for training and certification of the neural network. The 4mm×20mm×8mmx×mm four layers neural network model was established to predict the soldering forming. The prediction precision of network model was discussed. The results show that 4×20×8×l four layers neural network can be used for the prediction of soldering forming with high precision.
出处
《热加工工艺》
CSCD
北大核心
2013年第1期148-149,153,共3页
Hot Working Technology
基金
湖南省教育厅科学研究基金资助项目(11C0480)
关键词
神经网络
软钎焊
焊点成形
预测
neural network
soldering
weld forming
prediction
