Robust sensing and control of weld pool surface in gas tungsten arc welding and its modifications

Gas Tungsten Are Welding (GTAW) is an indispensable tool for metal joining due to its superior performance in root pass penetration. Although a number of researches have been focused on automated GTA welding, the bottleneck still lies in the need for an accurate and durable welding sensor. This doctoral research explores the effective and durable sensing systems for robust monitoring and control of the weld pool surface in GTA welding and its modifications. Among them is the Non-Transferred Plasma Charge Sensor (NTPCS), which takes advantage of the non-transferred plasma arc to measure the weld pool surface. Also utilized is the Efflux Plasma Charge Sensor (EPCS), which detects the penetration of workpiece from the back-side. A number of arc welding processes, including the Keyhole Plasma Arc Welding (K-PAW), the Soft Plasma Arc Welding (S-PAW), and the all-position pipe welding are addressed for sensing and control by the proposed sensors. For robust control of arc welding processes, which generally involve intensive disturbances, the interval model and interval model control algorithm are developed. Interval model envelops the process variations by the model parameters' bounds obtained via extreme-condition experiments. The prediction-based interval model control guarantees stability when tracking a fixed set point. The original linear interval model control is broadened by the nonlinear and adaptive versions for better performance in certain applications. Control experiments on the arc welding processes show that the proposed control systems are robust and quality welds are achievable.