Sensing and control of weld pool by fuzzy-neural network in robotic welding system

Abstract

It is important to control the penetration depth of the weld pool during welding, so as to obtain a good-quality weld, but it may be difficult to detect the penetration depth directly by using a visual sensor. In order to detect the penetration depth, the authors propose a penetration depth model based on a neural network. During welding, a fuzzy controller adjusts the welding current so as to obtain the desired penetration depth. Since the performance of the fuzzy controller depends on fuzzy variables, its tuning can be performed by using the neural network model. The validity of the fuzzy neural network is verified by some welding experiments.