Hybrid ANN-GA Modeling and Experimental Optimization of GMAW-Based Additive Manufacturing of Aluminum 5083 for Enhanced Mechanical Properties

Wire + Arc Additive Manufacturing (WAAM) based on Gas Metal Arc Welding (GMAW) has emerged as a cost-effective and high-deposition process for fabricating large-scale aluminum components. However, its application to non-heat-treatable aluminum 5083 remains limited by thermal-cycle in-stabilities, porosity, and non-uniform mechanical performance. This study presents an integrated experimental and artificial-intelligence framework for optimizing key GMAW parameters—welding current, wire-feed speed, and welding speed—to enhance the mechanical properties of WAAM-fabricated aluminum 5083 walls. An L9 Taguchi design was employed to quantify parameter effects, followed by analysis of variance (ANOVA) to identify dominant factors. Results indicated that weld-ing speed exerted the greatest influence on tensile strength (≈ 58.8 % contribution), whereas wire-feed speed and current primarily affected hardness through solidification behavior. An Artificial Neural Network (ANN) model was then developed to predict tensile strength and hardness with high accuracy (R > 0.99; MAPE < 1 %), demonstrating superior predictive performance over Taguchi and regression models. Integration of the trained ANN with a Genetic Algorithm (GA) enabled global optimization of process parameters, yielding an optimum set of 85.3 A current, 7.7 m/min wire-feed speed, and 3.8 mm/s welding speed, corresponding to predicted properties of 242.5 MPa tensile strength and 108.4 HV hardness. Experimental validation confirmed deviations below 1 %, verifying the model’s robustness. The proposed ANN–GA hybrid framework effectively captures nonlinear process–structure–property relationships, providing a reliable, data-driven pathway for achieving high-strength, defect-free aluminum components in WAAM and other additive manufacturing sys-tems.