Machine Learning-Driven Parametric Analysis of Eco-Friendly Ultrasonic Welding for AL6061-CU Alloy Joints

Abstract

This study outlines the research conducted to examine the mechanical behaviour and microstructural characteristics of Al-Cu dissimilar wires joints welded using ultrasonic joining process that commonly finds application in automotive components, heat exchangers and electrical home and industrial appliances. The primary focus is on the metallurgical transformations to evaluate the pattern of molecular diffusion and spread within the weld, the consistency of diffusion, and the resulting alterations in strength caused by ultrasonic vibrational heat. This procedure entails conducting experimental trials to join the wire materials according to per design of experiments, wherein the process parameters significantly influencing the output are systematically varied, and consequently, subjecting the joints to shear testing. Subsequently, the welded specimens undergo microscopic examination and the images are captured using image analysers. In addition, scanning electron microscopy (SEM) pictures are examined to gain insights into the surface shape and assess the degree of weld production and performance. The findings demonstrate a direct correlation between the vibrational temperature and the weld strength. In addition, the joint surface exhibits a consistent weld pattern in the majority of the samples, with just a few instances of inconsistencies when the trail is carried out at low heat input. Electrical resistance at the joints is measured to understand the electrical parametric variations if any due to process parameters. A machine learning tool is employed to forecast the weld strength and joint resistance for differing ranges of process parametric values and accordingly optimize it.