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。 Through the testing of the soldering process with different preheating temperature, chain speed, welding temperature and welding time, the average test data of the solder joint expansion rate obtained are used as the training samples and verification samples of the neural network, and a 4 × 20 × 8 × 1 four-layer neural network to predict the formation of solder joints, and to study the prediction accuracy of the network model. The results show that the four-layer neural network can be used to predict the formation of solder joints with high accuracy.