将人工神经网络技术引入焊接专家系统，建立了多种焊接规范参数设计网络模型。以实际焊接工艺数据对所建网络模型进行训练，经测试表明，网络具有很强的自学习自适应能力、容错与协调能力以及显著的联想功能。本文有效解决了焊接工艺设计专家系统的关键困难———焊接规范参数设计，为焊接专家系统研究开辟了新的方向。 The artificial neural network technology is introduced into the welding expert system, and a variety of welding specification parameter design network models are established. The actual welding process data is used to train the network model. The tests show that the network has strong self-learning adaptability, fault-tolerance and coordination ability and significant association function. This article effectively solves the welding process design expert system of the key difficulties - welding specifications of the design parameters for the welding expert system opened up a new direction.