Year of Publication


Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation




Electrical and Computer Engineering

First Advisor

Dr. Yuming Zhang


GTAW (Gas Tungsten Arc Welding) weld pool surface is believed to contain sufficient information to determine the weld penetration, from which skilled welders are able to control the welding process to desired penetration states. However, it is unclear how human welders extract the weld penetration from the observed weld pool surface. In this research, a novel method is studied to determine the weld penetration based on the dynamic change of the weld pool surface.

This study observes/measures/analyzes the development of a weld pool from partial to full penetration and correlates such change to the weld penetration. Similar trends in the weld pool surface are observed when the weld penetration changes from partial to full penetration despite the amperage used and material welded. Correlating the weld pool surface reflecting grayness and the development of the weld penetration from experiments shows: (1) the weld pool reflection intensity will increase while the weld penetration is increasing; (2) the increasing trends of weld pool reflection intensity will decrease when the full penetration is achieved; (3) the weld pool reflection intensity will increase after the full penetration is achieved. Such trend in the weld pool surface reflection intensity when the weld penetration develops is used as feedback signal to detect the weld pool penetration. To control the weld pool penetration, a first-order dynamic model is identified. Model Predictive Control (MPC) is used to control the weld penetration. Experiments verified the feasibility of this proposed method and established system.

Digital Object Identifier (DOI)