Friction stir welding (FSW) is the most widely used solid-state joining method for sheet and plate-like materials, valued for being a highly adaptable, environmentally sound, and energy-efficient process. The literature has repeatedly shown that FSW is a reliable joining technique for high-demand technological applications, particularly for high-strength aluminum and titanium alloys used in aerospace, which are difficult to join with traditional fusion welding methods. To describe the microstructural changes caused by solid-state FSW, many studies analyze mechanical parameters of FSW joints, including tensile strength, bending, torsion, elasticity, and fatigue responses. In recent years, the push to expand FSW's use, broaden the range of compatible alloy systems, and improve the resulting mechanical properties has led to significant advancements in this joining method. Accordingly, this review study provides a comprehensive investigation into the methodology and the materials relevant to this specific field.

This study investigated the dynamic behavior of AA7075 plates joined by Friction Stir Welding (FSW), focusing on the influence of key process parameters, rotation, and traverse speeds, on the resulting dynamic characteristics. Experimental Modal Analysis (EMA) was performed under free boundary conditions to determine resonance frequencies, mode shapes, and damping ratios, revealing that an increase in traverse speed consistently led to a decrease in natural frequencies across most modes, thereby indicating reduced joint stiffness attributed to insufficient heat input. Furthermore, localized weld defects caused significant damping variations, particularly in low-order modes. To complement the experimental findings and enable simultaneous, multi-output prediction of these coupled dynamic parameters, a Co-Active Neuro-Fuzzy Inference System (CANFIS) model was developed. The CANFIS architecture utilized spindle speed and feed rate as inputs to predict natural frequency and damping ratio for multiple vibration modes as tightly coupled outputs. The trained model demonstrated strong agreement and high predictive accuracy against the EMA experimental data, with convergence analysis confirming its stable learning and excellent generalization capability. The successful integration of EMA and CANFIS establishes a robust hybrid framework for both physical interpretation and intelligent, coupled prediction of the dynamic behavior of FSW-welded AA7075 plates.