Machine Learning-Based Design Parameters Recommendation In Construction Additive Manufacturing

Additive Manufacturing (AM) has gained attention in the building industry. However, leveraging this technology is challenging due to the sensitivity of AM processes to material properties and manufacturing configurations, which require precise calibration. Thus, we designed a Machine Learning (ML) model in this study to define material distribution parameters for a concrete wall segment based on numerically calculated thermal transmittance and material volume. For this, we executed the following steps: development of an algorithmic-parametric model of a single concrete wall; formation of a dataset based on design parameters and volume and thermal performance results; impact analysis of design parameters in performance; development of an ML model using the Extreme Gradient Boosting (XGBoost) algorithm; evaluation of the model's effectiveness. The results showed promising applications of ML in AM design optimization. We conclude by discussing ML's potential to assist AM design and address challenges from complex process parameters.