Development of a Real-time, Process Control Method Based on Neural Network Model Using Feedback of Weld Pool Geometric Parameters Measured by a Vision-based Technique and Experimental Verification for Automated Arc Welding Processes
The funded amount received from AICTE under Research Promotion Scheme (RPS) will be utilized to create facility with a vision system and a tri-axial manipulator in the Welding Research Lab. Research in the area of controlling the welding process with a feed-back of weld pool geometry has been of interest lately as a way to improve the quality and the consistency. In this study, weld pool geometry will be measured using a vision system for various sets of welding parameters and correlated to weld quality. Experimental data will be collected under various welding conditions such as current and travel speed for this purpose. A real-time controller based on neural network model to regulate welding parameters will be developed.