Accelerating extrusion-based additive manufacturing optimization processes with surrogate-based multi-fidelity models

Abstract

Fused deposition modeling (FDM) is by far the most common extrusion-based additive manufacturing technology. Affordability and feasibility promote the development of FDM technologies; nevertheless, product quality problems hinder the future growth of this advanced manufacturing technique. Optimizing the parameters of the manufacturing process can improve product quality. However, traditional optimization techniques require extensive experiments to determine the optimum condition. In this study, a low-fidelity numerical simulation predictive model and a high-fidelity experimental model were combined to iteratively optimize the additive manufacturing process. Although the proposed method was initially targeted for extrusion-based additive manufacturing processes, it was also verified with various practical additive manufacturing optimization problems. It is demonstrated that the proposed optimization algorithm outperformed traditional optimization algorithms by reducing the optimization cost by at least 14.6%. Moreover, the optimizer demonstrated superb noise tolerance ability.