DefinePK

Abstract

This paper presents a study on automatic welding speed control by monitoring of weld pool image using vision sensor to create a welding control system that able to produce good weld bead appearance and quality. The monitoring performs on autogenous Tungsten Inert Gas (TIG) welding with constant-current. In this experiment, welding torch set to be stationary while stainless steel SS-304 as work-piece placed on moving-table driven by electric stepper motor. CCD camera captured images of weld pool that formed during the welding process to be extracted by image processing algorithm. The pool geometry values that have extracted will be applied to the real time welding speed control system using Artificial Neural Network (ANN) system. The result showed that the system developed had succeeded to produce good weld bead appearance on several current values.