Compressive Strength Prediction of Nanosilica-Incorporated Cement Mixtures Using Adaptive Neuro-Fuzzy Inference System and Artificial Neural Network Models

Abstract

In recent years, through the development of nanoscience and technology, new ideas have emerged for enhancing the performance of cement composites. In this regard, nanomodified mixes, particularly those with nanosilica, have found a special position. However, there are challenges in using nanosilica in cement mixes, such as high price and workability problems. Thus, these materials must be consumed at certain levels to reach goal characteristics. In addition, there are complications in the properties and interactions of materials, which make it difficult to find a simple model for the prediction of concrete properties. In the present study, it has been tried to predict the compressive strength of cement composites utilizing artificial intelligent approaches, including an adaptive neuro-fuzzy inference system (ANFIS), artificial neural network (ANN) technique, and linear and nonlinear regression analyses. ANFIS and ANN are highly reliable methods for predicting the various properties of concrete; thus, these methods have been used extensively in concrete research. However, similar studies were not found on using these methods for prediction of compressive strength of cement mixtures with nanosilica. This study has utilized these methods to provide a comparison between the ANFIS and ANN models in predicting the strength of cementitious mixes and show the capability of the models of ANFIS and ANN compared with the traditional regression methods. For this purpose, the mix proportions and the quantity and size of nanosilica have been considered as input parameters, with the compressive strength of mortars as output parameters. The results indicate that ANN and ANFIS outperformed the regression analyses. Based on the obtained results, ANN had higher accuracy in predicting the compressive strength.