Global and Cluster-wise Regression Models for Total Bead Area as Welding Quality

Abstract

Abstract In recent years, the automated welding system is used in high volume production industries for efficient and precise welding works. It can be applied for not only a simple machine, but also production of a single part. Therefore, a new algorithm that estimates the optimal welding parameters on a given bead geometry and accomplishes the desired mechanical properties of the weldment in order to make the automatic GMA (Gas Metal Arc) welding process, desirable. The developed algorithm could use for a wide range of material thicknesses, and be applicable for all welding positions. In addition, the models must be available in the form of mathematical equations, which can program to the robot, and give a high degree of confidence in predicting the bead dimensions. In this study, two regression models employing global regression analysis and cluster-wise regression analysis are proposed to be applicable for prediction of optimal welding parameters on the total bead area. For development of the proposed regression models, an attempt has been employed for finding the optimal welding parameters. A full factorial design is adopted to investigate welding parameters effect on total bead area in lap-joint welding. as a function of key output parameters in the automated GMA welding process was utilized. Not only the fitting of these models were checked and compared by using a variance test (ANOVA), but also the prediction of total bead area using the developed regression models was carried out the basis of the additional experiments. The fitness and variance test of the regression model is performed by using experimental data and ANOVA. The prediction of total bead area was carried out via regression model. Experimental results showed that the fitting on the experimental data of the cluster-wise regression model with the RMS value of 0.0026 is better than the developed global regression model for determining total bead area.