Sensing of the weld pool depth with neural network and fuzzy control of seam tracking

Deals with the problem concerning the sensing of the weld pool and the tracking of the welding line with welding robots. In order to obtain a good quality of the welding result, it is important to control the weld pool depth and to keep the torch posture constant regardless of the external disturbance. First, a new method is proposed for sensing the weld pool depth with the neural network. The depth is kept constant regardless of the disturbance, such as the variation of the groove gap, with the fuzzy controller. Next, the method of seam tracking is discussed when the pipe is joined to the plane. The CCD camera and the laser are used to detect the welding line and to control the torch posture. The torch axis φ of the robot, its root axis θ, and the torch height are controlled with the fuzzy controller. The validity of the neural network and the fuzzy controller is verified by performing the experiments