Refined dimensional accuracy in FDM components via ensemble weight-optimized surrogates and hybrid NSGA-II-TOPSIS optimization

Abstract

In this paper, a novel approach was developed to predict the geometrical deviation and density of fused deposition modeling (FDM) parts and to optimize these outcomes by fine-tuning the relevant process variables. The prediction process utilized an ensemble approach, combining the optimized weighted sum of three individual surrogate models. These models correlate the input variables - nozzle temperature (NT), infill fraction (IF), and print speed (PS) - with the output responses of density, internal ovality (IO), and external ovality (EO). According to the results, the ensemble outperformed the individual surrogates, exhibiting enhanced predictive accuracy. Subsequently, non-dominated sorting genetic algorithm II (NSGA-II), a multi-objective optimization technique, was integrated with the technique for order of preference by similarity to ideal solution (TOPSIS), a multi-criteria decision-making method, to optimize the ensemble of surrogates (EoS) and generate a ranked set of Pareto-optimal solutions. The outcomes from the EoS, combined with the NSGA-II-TOPSIS hybrid optimization process, revealed that the most effective optimal control parameters were approximately 195 degrees C for NT, 50% for IF, and a range of 55-66 mm/min for PS. This hybrid approach affirmed the reliability and robustness of the identified processing parameters for the production of top-quality FDM parts with desired dimensional accuracy and density.