Machine learning based optimization of fly ash content for improving geopolymer concrete compressive strength.

Area of Science:

• Materials Science
• Civil Engineering
• Sustainable Construction

Background:

• Growing demand for sustainable construction materials.
• Fly ash-based geopolymer concrete (FA-GC) offers a promising alternative to traditional concrete.
• Need for accurate prediction and optimization of FA-GC properties.

Purpose of the Study:

• To predict the compressive strength (CS) of FA-GC using advanced AI models.
• To optimize FA-GC mix design for maximum performance.
• To identify key factors influencing FA-GC compressive strength.

Main Methods:

• Application of AI models: Tabular Prior-Data Fitted Network (TabPFN), Histogram-based Gradient Boosting (HistGBoost), M5Prime, and Automatic Feature Interaction Learning (AutoInt).
• Hyperparameter tuning using Optuna.
• Mix design optimization using metaheuristic algorithms: Harris Hawks Optimization (HHO), Grey Wolf Optimization (GWO), Lyrebird Optimization Algorithm (LOA), and Polar Bear Algorithm (PBA).
• Sensitivity analysis using SHAP values and partial dependence plots.

Main Results:

• TabPFN demonstrated superior predictive accuracy (R² = 0.981) and stability (lowest CI and R-Factor).
• HHO algorithm achieved the highest compressive strength of 61.56 MPa.
• Al₂O₃ (%) and SiO₂ (%) were identified as the most influential components for enhancing CS.

Conclusions:

• Advanced AI models combined with metaheuristic optimization significantly improve FA-GC prediction and mix design.
• This approach enhances the stability and reliability of geopolymer concrete solutions.
• Findings pave the way for more efficient and sustainable geopolymer concrete applications.