Presenting a novel approach for estimation the compressive strength of high strength concrete using ANN & GEP

In this article, the application of artificial neural networks in predicting the degree of concrete compressive strength of High Strength Concrete (HSC) was investigated. For this purpose, use was made of the pattern recognition neural network and the obtained data from the experimental tests for predicting the compressive strength degree of HSC. Five inputs from the HSC mix design were utilized for predicting the degree of compressive strength, by application of the scaled conjugate gradient backpropagation algorithm in neural network. The outputs were classified into 5 strength groups of M1, M2, M3, M4 and M5. The simulation results shows 97.9% accuracy in classifying the different predefined degrees of HSC using the confusion matrix diagram. Moreover, the cross-entropy error obtained from testing the neural network (NN) model and correlation coefficient (R2) of GEP for predicting compressive strength of the HSC were evaluated at 0.042096 and 0.9795, respectively, indicating high accuracy of the model. Application of this model could greatly help the persons, companies and research centers in terms of preparation and making of HSC with desired compressive strength, that are in need of this type of concrete.