Real-time control of welding penetration during robotic GTAW dynamical process by audio sensing of arc length

This paper presents a new technology of real-time arc length monitoring and sag-depression prediction through arc sound signal, which is essential to realize the welding penetration control during gas tungsten arc welding of arc length. A set of automatic measurement and control system have been proposed to achieve real-time arc length control via audio sensing system. After preprocessing of arc sound signal, the piecewise linear models of arc sound signal were established under two different arc length variation 3-4 and 4-5-6 mm, analyzing the prediction errors of linear model, which were proved to be good enough for online monitoring of arc length in pulse GTAW. Based on the linear relationship between arc sound and arc length, the linear fitting model was implemented on predicting the surface height of weld pool. A segmented self-adaptive PID controller was proposed to achieve the monitoring and controlling of arc length;, the confirmatory experiments have been designed to test the control effect of arc-length monitoring based on arc sound signal.