Model-based adaptive vision control of weld pool area

Weld pool geometry is a crucial factor in determining welding quality, especially in the case of sheet welding. Quality images that can be processed in real-time to detect the pool geometry are acquired by using a high shutter speed camera assisted with nitrogen laser as an illumination source. An adaptive system is developed to control the pool area using the vision feedback. A model is proposed based on an analysis of the timing logic of the sampling periods. Also, an online identification algorithm is developed to improve the estimation accuracy and convergence. Experiments show that the developed control system can overcome the influence of various disturbances (torch speed, heat transfer condition, initial parameters, and unexpected sensing failure) under practical welding conditions.