Prediction of concrete properties using machine learning algorithm

Abstract

This paper presents a machine learning model for the prediction of two mechanical properties, namely concrete compressive strength and chloride penetration depth. A total of five machine learning algorithms, including Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVR), Gradient Boosting (GB), and Artificial Neural Network (ANN), are employed on the collected data set. The winner model is then selected, and its hyperparameters are tuned using the particle swarm optimization (PSO) algorithm. To avoid overfitting of the model on the test data set 10 k-fold cross-validations are used. Features are handled for missing data set values using mean values of available data for the same features. The model performance results are measured using Root Mean Squared Error (RMSE) and coefficient of determination (R2). The proposed model yielded 97% accuracy with a 4 MPa value of RMSE, which indicates an efficient model. Out of all the models, Random Forest Regressor and Gradient Boosting Regressor Model performed well.