Prediction of Constituents of Concrete Mixtures Containing Fly Ash and Blast Furnace Slag Using Machine Learning Techniques

Abstract

The prediction of concrete mix proportions is of the utmost importance to civil engineers to complete the design process of structures. This process is usually done through a trial-and-error process which involves simple regression techniques and is usually done to achieve a specific strength at a specific age. The incorporation of supplementary cementitious materials into concrete mixtures for environmental purposes has deemed the prediction process more complex and created a need to come up with more advanced techniques. Furthermore, the ability to predict the constituents of concrete mixtures given multiple inputs is still limited. Hence, in this work several machine learning algorithms were utilized to make a prediction regarding mix proportions of concrete mixtures based on concrete compressive strength, concrete age, and density as inputs. Random forest, decision tree, and K-neighbors regressors were used to achieve this objective. Mean squared error as well as root squared error were used to measure the accuracy of the constructed models. Random Forest algorithm obtained the highest accuracy with 98.5%.