Machine Learning the Concrete Compressive Strength From Mixture Proportions

Abstract

Concrete mixture design usually requires labor-intensive and time-consuming work, which involves a significant amount of “trial batching” approaches. Recently, statistical and machine learning methods have demonstrated that a robust model might help reduce the experimental work greatly. Here, we develop the Gaussian process regression model to shed light on the relationship among the contents of cement, blast furnace slag, fly ash, water, superplasticizer, coarse aggregates, fine aggregates, and concrete compressive strength (CCS) at 28 days. A total of 399 concrete mixtures with CCS ranging from 8.54 MPa to 62.94 MPa are examined. The modeling approach is highly stable and accurate, achieving the correlation coefficient, mean absolute error, and root mean square error of 99.85%, 0.3769 (1.09% of the average experimental CCS), and 0.6755 (1.96% of the average experimental CCS), respectively. The model contributes to fast and low-cost CCS estimations.