Autonomous Decision-Making of Welding Position During Multipass GMAW With T-Joints: A Bayesian Network Approach

Abstract

This work investigates a Bayesian network model (BNM) to implement the autonomous decision-making process of welding positions in gas metal arc multipass welding with T-joints for automated manufacturing. The laser vision sensor is used to profile the weld seam, and the weld seam profile (WSP) is extracted with a novel scheme based on scale-invariant feature transform and orientation feature detection. The feature points of the extracted WSP are effectively identified through slope mutation detection. A BNM is built with these points and the determined welding state online, and the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">priori</i> probabilities of the model variables are acquired in a computational manner integrated with welding experience. The feature point with the maximum <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">posteriori</i> probability is selected as the current welding position using the evidence prepropagation importance sampling inference algorithm. The analytic hierarchy process and the C4.5 decision tree algorithm are used to compare with the proposed BNM regarding decision effectiveness. Experimental results show that the proposed BNM can give the effective decision-making results of welding positions with T-joints of different thicknesses and show great potential for higher manufacturing efficiency and automatic levels.