Dynamic nonlinear modeling of 3D weld pool surface in GTAW

• This paper has studied the application of computer vision technology in robotic GTAW.
• A 3d-vision sensing technology is used to rebuild the surface of the weld pool.
• The time-domain analysis of the system was done by conducting step responses.
• The dynamic relationship of current and pool geometric parameters is established.

As the weld pool contains abundant information about the welding process, the skilled welder always utilizes the observed information to control the weld pool to get a good weld quality. In this paper, a real-time vision sensing technology is used as the eyes of the welder to observe the gas tungsten arc welding (GTAW) pool surface and a 3D reconstruction technology is used to reconstruct the observed weld pool surface in three-dimensions. The geometric parameters are obtained from the reconstructed pool surface. Step responses are conducted to acquire the relationship between the weld pool geometric parameters and the random variation of current, and the time-domain analysis of the system is done. Various random experiments are also conducted and the dynamic relationship of the weld current and the weld pool geometric parameters is established by means of a non-linear Hammerstein model.