Prediction of strength for concrete specimens using Artificial Neural Networks

Artificial Neural Networks are trained for different types of concrete mixtures, in order to predict the 7-day and 28-day strength of concrete specimens. Both plain and admixture concretes are considered. Employing the Backpropagation algorithm, neural nets with one, two and three hidden layers are trained and compared. The most efficient networks are then selected and used for predicting the strength of concrete mixtures, with reasonably small errors.