Abstract
A visual sensing system has been established to monitor the weld pool in pulsed gas metal arc welding (P-GMAW). The system consists of a charge-coupled device (CCD), a frame-grabber and a composed filter system. To acquire quality images of the weld pool, the disturbance from the arc needs to be reduced. In this research, the arc light spectral distribution in GMAW is studied with a spectral analysis system and two suitable filtering spectral windows are indentified. As a result, two kinds of filter system are proposed. Studies have also been done in order to find the optimal timing and view angle for imaging. It is found that the end of the base current period in P-GMAW is an optimal instance for imaging and the weld pool can be best imaged from the upper rear in relation to the torch. Using the proposed filter, clear pool images have been acquired from the upper rear at the end of each base current period. Because of the satisfactory quality of the acquired images, image processing algorithms have been developed to extract the weld pool edge.