Revolutionizing Friction Stir Welding: A Synergy of Finite Element Simulation and Machine Learning for Enhanced Weld Quality

This talk delves into the transformative potential of integrating Finite Element Analysis (FEA) and machine learning (ML) for optimizing the Friction Stir Welding (FSW) process. A ۳D thermo-mechanical simulation of aluminum alloy (Al ۶۰۶۱) joints was developed using ABAQUS, accurately predicting nodal temperatures and residual stresses under varying process parameters. Nine cutting-edge ML regression models, including ensemble methods like Random Forest, Gradient Boosting, and XGBoost, were leveraged to predict and optimize welding outcomes with exceptional accuracy. The study demonstrates the critical impact of rotational speed and feed rate on heat generation and stress distribution, showcasing ML’s ability to enhance FSW performance. This research paves the way for real-time monitoring, adaptive control systems, and a new paradigm in high-performance welding technologies, offering a robust, scalable approach for superior welded joints. Join us to explore how computational intelligence is redefining the future of advanced welding processes.