Quality Monitoring Method Using Machine Learning for Resistance Spot Welding

Abstract

This study proposes and develops the proposed intelligent monitoring method for estimation, classification, and improvement of welding quality of resistance spot welding (RSW) using machine learning. In this paper, the authors focus on addressing the challenges in ensuring and improving welding quality with a proposed model (ie. multi-layer perceptron) for classification to determine the welding quality, and logistic regression aims to predict the welding current for improving the welding quality. In addition, the proposed model will be applied do developing the intelligent monitoring system based on analyzing the input gathered dateset through many practical experiments on RSW machine with AC inverter. The practical experiments are set up with Galvanized steel to collect the dataset aims to validate the performance of the proposed method.