Evolutionary AI-Based Algorithms for the Optimization of the Tensile Strength of Additively Manufactured Specimens

Abstract

In this research work, we investigated the optimization of input parameters for additively manufactured specimens fabricated by Fused Deposition Modeling (FDM) to maximize tensile strength. We employed two metaheuristic optimization algorithms, Particle Swarm Optimization (PSO) and Differential Evolution (DE), to determine the optimal input parameters, including infill percentage, layer height, print speed, and extrusion temperature. Additionally, we coupled both PSO and DE algorithms with the XGBoost algorithm, a powerful gradient boosting framework, to assess their performance based on Mean Squared Error (MSE) and Mean Absolute Error (MAE) values. Our study revealed that the MSE and MAE values of the coupled PSO-XGBoost algorithm were lower compared to the DE-XGBoost algorithm, indicating superior performance in finding optimal input parameters. The results suggest that the integration of PSO with XGBoost can provide an effective approach for optimizing FDM-fabricated specimens, leading to improved tensile strength and overall mechanical properties. This research offers valuable insights into the applicability of metaheuristic optimization algorithms in additive manufacturing and highlights the potential benefits of coupling these algorithms with machine learning models for enhanced parameter optimization. The findings contribute to the ongoing development of optimization techniques in additive manufacturing, providing a foundation for future work in this area.