Intelligent control of pulsed GTAW with filler metal

This paper addresses intelligent control of weld shape for plates with varied root openings during pulsed GTAW with filler metal, and is a development from the work in Refs. 1-3. The newly developed double-sided visual image sensing system could acquire the front topside, back topside, and backside images of the weld pool simultaneously in the same frame. The root opening and the double-sided weld pool geometry parameters were extracted online. A neural network model was established to predict the back-side width and topside height through welding parameters and topside shape parameters. Feasible intelligent control schemes were also investigated. In order to eliminate the effect of the root opening and stabilize both backside width and topside height, a double-variable controller was designed, in which the feedback control part regulates the pulse duty ratio to control backside width and, at the same time, the feed-forward control part adjusts the filler metal rate to achieve the desired topside height in order to compensate for the effects of the varying root openings.