Neuro‐fuzzy control of the weldpool in pulsed MIG welding

Abstract

Summary As part of current R&D work focused on developing the intelligence of arc welding robots, this paper deals with the problem of weldpool sensing and control. To obtain high‐quality welding, it is important to control the weldpool depth in robot welding regardless of any external disturbance, such as irregular groove gap. The method of controlling the weldpool depth without a mathematical model is discussed. Since it is difficult to measure the weldpool depth directly, it is estimated from the weldpool surface shape, groove gap, and welding current. A neural network is used to estimate the weldpool depth without a mathematical model. The weldpool depth is controlled from the output of the neural network using the fuzzy controller. Neural network and fuzzy controller application is validated in welding experiments.