Visual sensing of the weld pool geometry from the topside view in keyhole plasma arc welding

Weld pool and keyhole behaviors are the key factors to determine the process stability and weld joint quality in keyhole plasma arc welding. In this study visual sensing of the topside weld pool based on a low-cost CCD camera was realized. A visual sensing system with special dimmer glass and bandpass filtering was developed for solving the problems like strong arc disturbances and large radiation intensity gradient. Boundaries of the complete topside weld pool were acquired through image processing of the captured raw images. Weld pool dimensions were then calculated and calibrated. The relationships between the dimensions and welding current, welding speed and plasma gas flow rate were investigated. Three stages, including blind keyhole, unsteady keyhole and steady keyhole, were defined to describe the weld pool behaviors along with the keyhole’s evolution. The variation trends of weld pool shapes were discussed to study the possible mechanism of the complex physical processes. For keyhole plasma arc welding, this provides the industry a prospective low-cost way to predict the penetrations by sensing weld pool and keyhole from topside.