ABSTRACT
Friction stir welding (FSW) is a technique used to join materials that fusion welding deems difficult. FSW is largely a solid-phase technology that has been shown to be effective since it can produce inexpensive, low-distortion welding. Analyzing the heat transfer can be used to estimate the stresses and welding efficiency. With the aid of industrial equipment, sheets of dissimilar/similar Ti-6Al-4V of various thicknesses were bonded using the FSW process. In FSW, the creation of strong couplings between materials is crucial. Selecting the right welding conditions to create sound joints is one of the trickiest difficulties. Traditionally, welding parameters were established through a time-consuming process of trial-and-error development. Additionally, because welds can be manufactured with a wide range of optimal welding parameters, an optimized combination of welding parameters was not attained. To attain better performance in industrial applications, the major production parameters for welding were carefully discussed in detail, along with the joints’ microstructure, hardness, and residual stress. Numerous researchers have recently created algorithms to improve industrial processes. The use of these machine learning (ML) approaches in FSW enables it to anticipate the fault before it manifests. To forecast the error proportion for the FSW technology, ML techniques including multi-linear regression, K-nearest neighbor, random forest, adaptive neurofuzzy interference system, regression model, support vector machine, and artificial neural networks were discussed to regulate microstructural changes, fracture failure, and detect flaws. This chapter looks at ML applications in FSW compared with the conventional optimization tool.
