Friction stir welded joint strength prediction with advanced data-driven modeling by using machine learning techniques

This study focuses on predicting the ultimate tensile strength (UTS) of friction stir welded dissimilar aluminium alloys using data-driven machine learning approaches. Two predictive models, support vector machine (SVM) and artificial neural network (ANN) were created to model the correlation between important process parameters tool rotational speed, welding speed, tool geometry (D/d ratio) and tilt angle and the achieved joint strength. Experimental data were collected, pre-processed, and used to train and evaluate the models. The results show that SVM outperforms ANN in accurately predicting UTS, with higher correlation and lower error metrics. Feature importance analysis revealed that tool rotational speed and D/d ratio are the most influential parameters affecting weld strength. Incorporating predictive modelling in FSW can significantly improve process efficiency, reduce trial-and-error experimentation, and enhance the quality of welded joints.