Friction stir welding process has been studied extensively in the last decades since its early stage. Most of the research done so far is related to the process development including tool design, material weldability, post-weld mechanical behavior, and microstructural properties. More recently, in-line process monitoring and artificial intelligence algorithms are introduced into this process, but mainly to specific material configuration and joint thicknesses. This study will focus on the evaluation of different machine learning approaches including principle component analysis, K-nearest neighbor, multilayer perceptron, single vector machine, and random forest methods on a friction stir welding cell environment. The input variables provided from this cell environment are namely divided into two groups: one group refers to the application variables and the other group is related to the friction stir welding process variables. The application variables target the aluminum alloys, joint configuration, sheet thicknesses, initial mechanical properties, and their chemical composition. The friction stir welding process variables dictate the rotational speed, travel speed, forging force, longitudinal and transverse forces, torque, and specific energy. The output response to model from these machine learning algorithms is the defect index, which has been quantified using high-resolution immersed bath ultrasounds. This nondestructive evaluation technique has been described previously, which can detect defects ≥150 µm in thin sheets. The defect index has been classified into five classes, which is distinguished by the nature of defect, cold weld, or hot weld, as well as the width of the internal volumetric defect upon ultrasound C-scan result. The dataset, which is composed of around 500 various process conditions, has been generated over the last few years and the variables were taken exclusively in constant weld regime and in the force control mode using the output average values. This paper compares the best resulting machine learning methods applied on a friction stir welding cell basis, which is the K-nearest neighbor and multilayer perceptron algorithms. The K-nearest neighbor model reaches a deviation of 0.55 on the defect index in comparison with the experimental values, which is slightly better than the multilayer perceptron model, which obtains a score of 0.69. Over the initial 59 available model parameters, 10 and 15 of them were retained in the final algorithm using these techniques. The main predictors include the material thickness, base material ultimate tensile stress, rotational speed, travel speed, weld forces, and specific energy. The K-nearest neighbor model was able to provide a map of defect indices with regard to rotational speed and travel speed but was only possible when a higher density of data was found within the prediction area. A data density score was also included within the model to inform the end-user about the prediction reliability. The machine learning models are mainly about differentiating various cases rather than representing the physical phenomena as determined using the finite element analysis. That being said, in order to improve the prediction reliability as well as the machine learning models, the data twinning concept, which consists of generating simulated friction stir welding process conditions by finite element analysis, is briefly discussed.

The growing implementation of aluminum alloys in industry has focused interest on studying transformation processes such as laser welding. This process generates different kinds of signals that can be monitored and used to evaluate it and make a quality analysis of the final product. Internal defects that are difficult to detect, such as porosity, are one of the most critical irregularities in laser welding. This kind of defect may result in a critical failure of the manufactured goods, affecting the final user. In this research, a porosity prediction method using a high-speed camera monitoring system and machine learning (ML) algorithms is proposed and studied to find the most performant methodology to resolve the prediction problem. The methodology includes feature extraction by high-speed X-ray analysis, feature engineering and selection, imbalance treatment, and the evaluation of the ML algorithms by metrics such as accuracy, AUC (area under the curve), and F1. As a result, it was found that the best ML algorithm for porosity prediction in the proposed setup is Random Forest with a 0.83 AUC and 75% accuracy, 0.75 in the F1 score for no porosity, and 0.76 in the F1 score for porosity. The results of the proposed model and methodology indicate that they could be implemented in industrial applications for enhancing the final product quality for welded plates, reducing process waste and product quality analysis time, and increasing the operational performance of the process.

Friction stir lap welding of 1.5-mm thick 6082-T6 aluminum alloy to 2-mm thick galvanized DP800 steel (Zn-coated) was carried out. Optimal welding conditions were obtained aiming to defect-free joints with good mechanical properties. The interfacial intermetallic compounds (IMCs) at the stir zone and hook zone were characterized under different revolutionary pitches. With a revolutionary pitch of 1.0 mm/rev, maximum joint strength reached 71% of that of the aluminum alloy. In the meantime, the average thickness of IMC layer is less than 1 μm; Al_3.2Fe in the Al-rich side and Al₅Fe₂ in the Fe-rich side at the interfaces of stir zone while Al₆Fe and nanocrystalline close to Al_3.2Fe at the interface of the hook zone. At a relatively lower revolutionary pitch (0.5 mm/rev), Zn was found with the aggregation of Si and Mn at the hook-zone interface, leading to the generation of Al-Fe-Si phase thus decreasing the thickness of the IMC layer. In the stir zone, the revolutionary pitch has a significant influence on the interfacial microstructures. The interfacial IMC layer at 1.0 mm/rev is simple and flat, but the one at 0.5 mm/rev becomes thicker and more complex. Stir zone aluminum under different revolutionary pitches is similar in microhardness and tensile behavior. The mechanical response of joints was modeled based on linear mixture law with an iso-strain assumption and neglection of the IMC layer. The modeling results are in good agreement with the experimental ones indicating the resultant interfaces act as good as the good boundaries between dissimilar Al/steel joints.

AA7075 is considered a ‘non-weldable’ alloy using fusion welding methods. In this study, laser welding is applied in pulse mode to weld 2 mm thick AA7075 aluminum alloy plates using different fillers. The aim is to identify the influence of welding parameters and fillers on solidification cracking susceptibility during laser welding using the circular patch test (CPT). X-ray radiography was used to detect and measure cracks in the CPT samples. Furthermore, the effects of the laser welding process and chemical composition of fillers on the accumulated crack length (CCL), microstructure, and mechanical properties were investigated. Moreover, the mechanical behavior and local deformation of the fusion zone (FZ) were investigated using micro-flat tensile tests with digital image correlation. The mechanical properties of the FZ were correlated with the CCL as well as with the microstructure of the FZ, which was investigated experimentally. The results show that the chemical composition of fillers and welding speed affect the CCL of solidification cracks. Changes in the microstructure were observed within the fusion zone, and the structure became uniform and finer with the formation of Mg₂Si and magnesium-rich, copper, and zinc (η-phase) particles.