Performance Comparison of Machine Learning Models for Concrete Compressive Strength Prediction

Area of Science:

• Civil Engineering
• Materials Science
• Computational Science

Background:

• Traditional concrete compressive strength testing is time-consuming and complex.
• Accurate strength prediction is crucial for construction safety and efficiency.
• Machine learning offers potential for faster and more reliable predictions.

Purpose of the Study:

• To evaluate and compare the performance of various machine learning models for predicting concrete compressive strength.
• To identify the most accurate and efficient model for practical application in the construction industry.

Main Methods:

• Four machine learning models were employed: artificial neural network (ANN), multiple linear regression, support vector machine, and regression tree.
• A dataset of 1030 concrete samples was preprocessed and split into training (70%) and testing (30%) sets.
• Performance was evaluated using metrics like mean absolute deviation, root mean square error, coefficient of correlation, and mean absolute percentage error. The ANN model included further validation (15%) and testing (15%) subsets.

Main Results:

• The artificial neural network (ANN) model demonstrated superior performance compared to multiple linear regression, support vector machine, and regression tree.
• The ANN model achieved higher accuracy and efficiency in predicting concrete compressive strength.
• Evaluation metrics confirmed the ANN's predictive capabilities.

Conclusions:

• Machine learning, particularly the ANN model, provides a streamlined and efficient method for determining concrete compressive strength.
• This approach can significantly reduce the time and complexity associated with conventional testing methods.
• The findings support the adoption of ML models in the construction industry for enhanced material assessment.