Optimizing and hyper-tuning machine learning models for the water absorption of eggshell and glass-based cementitious...

Area of Science:

• Materials Science
• Sustainable Construction
• Machine Learning Applications

Background:

• Cementitious composites degrade due to water-facilitated ion transport, necessitating enhanced resilience.
• Repurposing waste materials like eggshell and glass powder promotes eco-friendly construction.
• Understanding water absorption is crucial for improving the durability of sustainable cementitious materials.

Purpose of the Study:

• To investigate the impact of eggshell and glass powder as partial replacements for cement and sand on mortar's water absorption capacity.
• To develop and validate machine learning (ML) models for predicting water absorption in these eco-friendly composites.
• To assess the influence of constituent materials on water absorption through sensitivity analysis.

Main Methods:

• Utilized Gene Expression Programming (GEP) and Multi Expression Programming (MEP) for ML modeling.
• Performed sensitivity analysis and interaction research to determine input variable importance.
• Validated ML model predictions against experimental data, achieving high R2 values (0.90 for MEP, 0.88 for GEP).

Main Results:

• Developed accurate ML models (GEP and MEP) for predicting water absorption in cementitious composites.
• Sensitivity analysis identified eggshell powder, sand, and glass powder as key factors influencing absorption capacity.
• The models provide unique mathematical formulas for predicting material properties.

Conclusions:

• The developed ML models accurately estimate water absorption in mortars incorporating eggshell and glass powder.
• These models offer a valuable tool for scientists and engineers to assess and optimize mixture proportions, saving time and resources.
• The study demonstrates the potential of ML in advancing the development of sustainable and resilient construction materials.