Physics-informed modeling and process optimization of friction stir welding of AA7075-T6 with a zinc interlayer

Abstract

Friction Stir Welding (FSW) is a solid-state joining method commonly used for joining both similar and dissimilar high-strength, low-melting-point alloys like AA7075-T6. However, the conventional FSW of AA7075-T6 continues to face challenges, including inadequate joint strength, poor interfacial bonding due to inadequate wettability and diffusion, corrosion susceptibility, non-uniform heat distribution, and defects. This study is the first to combine a zinc interlayer with machine learning (ML) based optimization in the FSW of AA7075-T6. Artificial Neural Networks (ANN), Support Vector Regression (SVR), Random Forest Regression (RFR), a Genetic Algorithm (GA) for optimization, and Response Surface Methodology (RSM) for statistical modeling were used to analyze a dataset of 60 observations. The models that are included in the hybrid framework of ANN, SVR, and RFR have all demonstrated noteworthy prediction strengths. The optimum FSW parameters were shown to be: a tool speed of 600 rpm, a pin radius of 5.71 mm, a shoulder radius of 20 mm, and a plunge force of 6369.48 N, with a predicted peak temperature value of 675.71 K. The ANN model yielded an extremely low prediction error of 0.973 %, while the validation through FEA showed an accuracy with only 1.79 % deviation. The efficiency of this framework in optimizing the FSW of AA7075-T6 was confirmed by the significant improvement in thermal performance caused by the zinc interlayer.