AI-driven Optimization of 3D Print Parameters

Abstract:

The quality and efficiency of Fused Deposition Modelling (FDM) 3D printing are highly dependent on the careful selection of process parameters such as layer height, infill density, print speed, and nozzle temperature. The standard parameter tuning processes are mostly rule-based and time-consuming, frequently involving much trial-and-error process. The paper proposes an AI-based framework of multi-objective optimization of 3D printing conditions, which is based on machine learning and evolutionary method. It was produced systematically as a result of experimentation, and it reflected the relationships between process control parameters and the most important measures of performance such as tensile strength, surface roughness, and the time it takes to print. Supervised learning algorithms, especially Random Forest and Boost, were able to predict very well (R 2 > 0.85). Integration of these models with NSGA-II was used to find Pareto-optimal sets of parameters that showed trade-offs between the performance and efficiency. The experimental verification revealed that AI-optimized settings demonstrate good results in performance compared to the default settings provided in a slicer, being more than 20% stronger, providing an improvement in finish and speed. The study has a contribution to the intelligent additive manufacturing, as it allows the controlled tuning of parameters automatically, precisely, and at scale.