Machine learning approaches for forecasting compressive strength of high-strength concrete

Area of Science:

• Civil Engineering
• Materials Science
• Computational Science

Background:

• Compressive strength is a critical mechanical property of High Strength Concrete (HSC), essential for structural safety.
• Traditional laboratory methods for determining HSC compressive strength are resource-intensive, involving significant time and cost.
• Artificial intelligence (AI) presents a promising avenue for developing efficient and accurate predictive models.

Purpose of the Study:

• To develop and evaluate a machine learning (ML) model for predicting the compressive strength of HSC.
• To investigate the efficacy of various regression models in predicting HSC compressive strength using Python.
• To demonstrate the potential of AI in optimizing concrete material design and performance assessment.

Main Methods:

• Utilized a dataset derived from original experimental tests on HSC.
• Employed Python programming language for developing and implementing ML models.
• Investigated multiple regression models, optimizing hyperparameters and evaluating performance using metrics like MAE, MSE, and R-squared.

Main Results:

• The XGBoost model achieved a high R-squared value of approximately 0.94, indicating superior predictive accuracy.
• Hyperparameter tuning was crucial for optimizing model performance.
• The study successfully demonstrated the capability of ML models to predict HSC compressive strength accurately.

Conclusions:

• Machine learning models, particularly XGBoost, show significant potential for accurately predicting HSC compressive strength.
• Python is a suitable programming language for developing reliable predictive models in civil engineering applications.
• AI-driven approaches can substantially reduce the time and cost associated with material property testing.