Dynamic control of 3D weld pool surface based on human response model

Skilled human welders are so far the only to be capable of producing quality welds against perturbations. They adjust welding parameters based on their observation on 3D weld pool surface. Modelling their response to the 3D pool surface thus leads to next generation intelligent welding robots without human physical limitations. This paper first models this response as a human intelligence based controller. Its controlled process, with 3D pool surface as output and welding parameters as input in gas tungsten arc welding (GTAW), is then Hammerstein modelled in order to analyse its effectiveness. Analysis shows that this controller robustly controls the 3D pool surface under various perturbations. Closed-loop control welding experiments further validated this effectiveness. A foundation is thus established to rapidly transform skilled human welder's intelligence into robotic welding systems.