Experimental and theoretical analysis of the welding process parameters for UTS with different methods

Abstract

Over past decades, there have been a wonder growth in interest in information systems technology and the application of soft computing (SC) techniques to engineering systems. SC is not a single methodology, but it is a partnership and an innovative approach to constructing computationally intelligent systems. Artificial Intelligence and SC techniques are widely used to solve real world problems. In the current investigation, the optical and scanning electron microscopes (SEM) were used for microstructural studies. The welding process parameters (WPPs) were considered key aspects in the determination of the ultimate tensile strength (UTS) of the joints. The results were obtained experimentally. In order to investigate the effect of the WPPs on the UTS, the artificial neural network (ANN), Taguchi and the adaptive-neuro interference fuzzy system (ANFIS) approaches were developed and compared utilizing experimental results. It was found that the ANFIS provides better results than the ANN and Taguchi but the explicit formulation cannot be obtained analytically for the system. It can be concluded by comparing the three approaches that the ANN has a mathematical formulation with 95.53% accuracy and mean absolute percentage error of 4.47% and can be used to estimate the UTS of the joints for next studies. It was also noted that all WPPs had an important effect on the UTS of the joints.