Cement production is one of the important sources of global CO ₂ emissions, and the development of sustainable concrete materials is of great significance for reducing carbon emissions.
Floating stone concrete can improve its durability while reducing environmental burden by adding floating stones to replace some of the cement used.
However, the mix design of pumice concrete involves the interaction of multiple materials, and predicting the 28 day compressive strength faces challenges.
Traditional empirical formulas have limited accuracy, and although machine learning models can improve prediction accuracy, a single algorithm has shortcomings in generalization ability.
This study constructed a hybrid ensemble learning model based on grey wolf optimizer to achieve high-precision prediction of compressive strength of pumice concrete, and provided interpretable engineering insights through SHAP analysis.
.