Using Neural Networks for Prediction of Properties of Polymer Concrete with Fly Ash

Abstract

This paper presents the results of studies conducted with neural networks on determining the properties of polymer concrete with fly ash. Polymer concrete with different contents of fly ash and resin was prepared and tested for determining the influence of fly ash on the properties. Using neural networks, the experimental results were analyzed for predicting the compressive strength and flexural strength, and also on the basis of a model with given values of properties, to ascertain the composition (content of resin, aggregate, and fly ash). Eleven sets were considered for training and four for verification. Reverse modeling proves that the largest values for compressive strength and flexural strength are obtained for a resin content of approximately 15–16%, and a fly ash content of approximately 8–9%.