Prediction Of Concrete Compressive Strength Using Machine Learning Algorithms

Abstract

This study investigates the use of machine learning techniques to predict the compressive strength of concrete using key input parameters such as water-cement ratio, cement content, and aggregate size. Several models, including Support Vector Regression (SVR) and Backpropagation Neural Network (BPNN), were applied to a comprehensive dataset of concrete mixtures. The results demonstrate that both models exhibit high predictive accuracy, with the SVR model achieving an R² score of 0.877 for the test set and the BPNN model scoring 0.872. Feature importance analysis revealed that the water-cement ratio and cement content were the most influential factors in determining compressive strength. The findings highlight the potential of machine learning in optimizing concrete mix designs, offering a more accurate prediction than traditional empirical methods. Future research should focus on incorporating additional variables and advanced optimization techniques to further improve the predictive capabilities of these models.