Comprehensive Machine Learning-Based Model for Predicting Compressive Strength of Ready-Mix Concrete.

Area of Science:

• Civil Engineering
• Materials Science
• Data Science

Background:

• Concrete compressive strength (CS) is crucial for structural safety, but traditional testing is time-consuming and expensive.
• Numerous factors influence concrete CS, making accurate prediction challenging.

Purpose of the Study:

• To develop an efficient machine learning framework for predicting concrete CS.
• To identify and utilize the most influential factors affecting concrete CS.
• To reduce the cost and time associated with concrete CS testing.

Main Methods:

• Systematic categorization of influential factors into the 4M1E framework (man, machine, material, method, environment).
• Application of a genetic algorithm (GA) for feature selection to identify key CS predictors.
• Utilization of random forest (RF) as the machine learning algorithm for CS prediction.

Main Results:

• The GA effectively identified the most significant factors influencing concrete CS.
• The RF model demonstrated high accuracy in predicting CS, evidenced by a Pearson correlation coefficient of 0.9821.
• The model achieved low mean absolute percentage error (0.0394) and delta (0.395), indicating reliable predictions.

Conclusions:

• The proposed machine learning framework provides an accurate and reliable method for evaluating concrete CS.
• This approach offers a cost-effective and time-efficient alternative to traditional CS testing methods.
• The findings support the use of data-driven models in civil engineering for material property assessment.