Abstract: This paper investigates practical application of intelligent control of the pulsed GTAW pool dynamic process by fuzzy logic inference and neural networks. Firstly, the neural network models for the flat butt joint pool character in pulsed GTAW dynamic process is established, such as the maximum width, length, area parameter models. And based on the experimental data, the fuzzy control rules for the welding process are built up by fuzzy identification algorithm. And then, the fuzzyneural network controller with self-learning and adaptive ability is designed for the welding process. The experiment on the process shows that the fuzzy-neural network controller has effected some intelligent control results.