Hybrid machine learning model using gorilla troops optimizer for accurate prediction of interlayer bonding strength in 3D-printed concrete

Abstract

Interlayer bonding strength (IBS) plays a pivotal role in 3D-printed concrete (3DPC), emphasizing the need for a reliable predictive model. A new hybrid model is proposed in this study, which leverages the gorilla troops optimizer (GTO) to fine-tune the hyperparameters of the extreme gradient boosting (XGBoost) algorithm. For comparison, XGBoost and decision tree (DT) models were also developed. To enhance interpretability, SHapley Additive exPlanations (SHAP) were employed to highlight the most influential factors affecting IBS. The proposed GTO-XGBoost model outperformed the other approaches, attaining a correlation coefficient of 0.974, compared to 0.958 for XGBoost and 0.930 for DT. The findings demonstrate that the GTO-XGBoost model offers a reliable solution for predicting IBS, contributing to the advancement of 3D printing technology in construction.