Recommendation System with Artificial Intelligence for Welding Quality Improvement

Abstract

This paper proposes the recommendation system (RS) to support the welding quality improvement system (WQIS) with artificial intelligence (AI). The proposed RS is applied to make suggestions for beginners and experienced workers about the welding quality predictions based on weld button size and the welding quality improvement by increasing welding current. The goal of the paper aims to develop an RS that has the ability to learn, analyze, predict and make these suggestions to humans through AI. Support Vector Machines (SVM) have been employed to predict the welding quality with impact parameters such as instantaneous (IHR), electrode tip diameter (De), and the status of welding current (Iw ) during the operation of the resistance spot welding (RSW) machine. The practical experiments are set up with an RSW machine using an AC inverter on Galvanized (GI) steel to collect the dataset for the SVM model. Through the experimental result, the effectiveness of the utility application is validated. In addition, these experimental results should be helpful for developing the high-performance RSW machine with AI applications in practice.