Machine Learning Techniques for Evaluating Concrete Strength with Waste Marble Powder.

Area of Science:

• Materials Science
• Civil Engineering
• Computational Intelligence

Background:

• Concrete strength is crucial for construction.
• Waste marble powder presents disposal challenges.
• Utilizing waste marble powder can enhance sustainability in concrete production.

Purpose of the Study:

• To predict the compressive and flexural strengths of concrete mixes.
• To evaluate the efficacy of various machine learning models in predicting concrete properties.
• To assess the use of waste marble powder as a sustainable alternative in concrete.

Main Methods:

• Experimental data from laboratory tests on concrete mixes.
• Application of Support Vector Machines (SVM), SVM with bagging, Stochastic, Linear Regression, and Gaussian Processes (GP).
• Statistical criteria were used to evaluate model performance.

Main Results:

• Gaussian Process (GP) and Support Vector Machines (SVM) demonstrated high accuracy in predicting flexural and compressive strengths, respectively.
• GP and SVM Stochastic models showed superior performance with high correlation coefficients (0.8235 and 0.9462) and low error metrics.
• Curing days, aggregate type, cement content, water content, and marble powder content were identified as key predictors.

Conclusions:

• GP and SVM are reliable methods for predicting concrete strength.
• These computational models can significantly reduce experimental workload and time.
• Waste marble powder can be effectively utilized as a partial replacement for cement and sand in concrete mixes.