Closed-loop control of weld penetration using front-face vision sensing

Feedback control of weld penetration based on a front-face sensor is a challenging problem in the field of welding. A novel vision-based approach is proposed in the paper for full penetration control of the gas tungsten arc welding (GTAW). Owing to the relationship between the front-face weld geometry and back-face weld width (representation of the full penetration state), which has been reported earlier by the present authors, it is possible to use the front-face geometry as feedback of full penetration. Based on the dynamic modelling and the analysis of accepted adaptive control algorithms, an adaptive predictive decoupling controller is developed. Simulations and experiments under a variety of welding conditions have been conducted to verify the effectiveness of the proposed approach and controller.