Abstract: A self-learning fuzzy neural network control approach to the dynamic process of welding current-top bead width in the pulse TIG welding is presented. Using artificial neural networks for modelling the process, the fuzzy controller described in the analysis formula with control error, error change and error accelation is real-timely regulated by self-learning weight factors. The results of experiment and simulation on the pulse TIG welding process show that the self-learning fuzzy neural network control scheme presented in this paper is effective.

Key words: Control bead width, Fuzzy control, Neural network, Pulse TIG welding, Self-learning