ANN and GEP prediction for simultaneous effect of nano and micro silica on the compressive and flexural strength of cement mortar

Abstract

Artificial neural networks (ANNs) and genetic expression programming (GEP) have recently been used to model the properties of cement mortar containing nano silica (NS) or micro silica (MS) to minimize the experimental work. Appropriate ANN and GEP models were proposed in this paper to predict the simultaneous effect of NS and MS on cement mortar properties because available models are not suitable to predict simultaneous effect of NS and MS. For this purpose, ANN and GEP models were trained on 640 different mixture proportions considered from literature and the prediction results were compared to available models. Moreover, in order to validate the proposed ANN and GEP models, a total of 480 compressive (50 × 50 × 50 mm) as well as 96 flexural (40 × 40 × 160 mm) specimens were constructed while considering cement strength class of 42.5 MPa and also, mechanical properties were tested.The higher accuracy results observed when in proposed models for predicting of compressive strength of cement mortar used normalized input parameters. Moreover, flexural strength of cement mortar estimated by the new ANN and GEP models is validated by experimental effort and the results show high accuracy compared to previous models. Finally, a comparative evaluation was also performed on two conditions of with and without considering porosity as an input parameter in the new suggested ANN and GEP models.