Friction Stir Welding Dissimilar AA-5083 and AA-6061 Alloys with Fuzzy Logic and Enhanced MLP Predictive Model

The joining of dissimilar aluminum alloys, such as AA-5083 and AA-6061, via Friction Stir Welding (FSW) presents significant challenges due to differences in their material properties. This study proposes a predictive model combining Fuzzy Logic and an Enhanced Multi-Layer Perceptron (MLP) neural network to optimize the FSW process for these alloys. The model aims to predict key welding parameters, such as tensile strength and hardness, based on input variables including tool rotational speed, welding speed, and plunge depth. The Fuzzy Logic system captures the non-linear and uncertain relationships between process parameters, while the Enhanced MLP refines these predictions through a robust learning process. Experimental data was collected for various welding conditions, and the model's performance was evaluated against actual results. The proposed model demonstrated high accuracy in predicting the mechanical properties of the welded joints, providing valuable insights for improving the quality and reliability of FSW joints in dissimilar AA-5083 and AA-6061 alloys. This approach can be used to optimize welding parameters in real-world applications, offering a tool for manufacturers to achieve superior weld quality while minimizing defects.