Table 1 Summary of the literature review.

FSW process

Singh et al.¹; Gebreamlak et al.²; Kaygusuz et al.³; Mishra and Ma⁵; Khalafe et al.¹³; Sezhian et al.⁶; Kwon et al.⁸

Provides a fundamental understanding of FSW required for RQ1 and RQ3

FSW is a solid-state joining process that has several advantages, including high bond strength and minimal defects

Improves mechanical properties in alloys such as AA 1100

A lack of real-time models for industrial applications

There is little data on long-term reliability and challenges in welding dissimilar materials

Parametric study in FSW

Gite et al.⁷; Kwon et al.⁸; Chauhan and Kumar⁹; Prakash et al.¹⁰; Albunduqee and Al-Bugharbee¹¹; Rani et al.¹²; Hasnol¹⁴; Khalafe et al.¹³,

Directly related to RQ1 by optimising FSW parameters for reliability in AA 1100 alloy plates

Identifying key parameters (rotational speed, welding speed, tool geometry)

Optimising parameters produces high-quality, defect-free joints

Few studies on dissimilar materials and parameter interactions to reduce defects

There is a need for refined tool designs and advanced optimisation

Parametric optimisation techniques

Osarumwense and Rose¹⁵; Ghetiya et al.¹⁶; He et al.¹⁷ Azadegan et al.¹⁸; Chi and Hsu¹⁹; Prabhakar et al.²⁰; Isa et al.²¹; Alok et al.²³; Toorajipour et al.²⁵

Relevant to RQs 1 and 3 for determining optimal welding parameters and evaluating reliability using methods such as Weibull analysis

Optimisation techniques such as Taguchi, RSM, and bio-inspired algorithms improve FSW results

Soft computing techniques improve decision-making

Optimisation for materials with different properties is still in its early stages

Advanced bio-inspired algorithms are required for dynamic control, as is integration with AI

Soft computing and AI in FSW

Okuyucu and Arcaklioglu²⁷; Boulahem et al.²⁸; Lomolino et al.³³; Elanchezhian et al.³⁴; Gokulachandran and Mohandas³⁶; Rizvi et al.³⁷; Pitchipoo et al.³⁸; Ilangovan et al.³⁵; Cederqvist and Öberg³⁹

Addresses RQ2 by comparing ANN to other AI models and RQ3 through AI-based reliability evaluations

AI models such as ANN predict weld quality and optimise parameters

Soft computing methods outperform traditional models for predicting weld results

The AI models for real-time control have not been fully implemented

More research is needed into AI-based long-term reliability and integration with real-time feedback systems