Weld Pool Surface Model Establishment for GTAW Based on 3D Reconstruction Technology

摘要

In Gas Tungsten Arc Welding (GTAW) process, weld pool contains information that can be used to establish process model, realize desired penetration control and study the weld pool characteristic. Hence, various sensing methods are used to sense weld pool status. At the same time, many researches are conducted to establish the weld process model, and obtain desired weld process control. In this paper, a three dimensional weld pool sensing system was introduced, and the image processing, weld pool reconstruction results were presented. Then, based on the reconstructed 3D weld pool geometry, an artificial neural network (ANN) model was established to describe the weld pool geometry. Simulation results show that the ANN model can reflect weld pool geometry accurately, which will benefit the GTAW process analysis and control.