Estimation of compressive strength of diatomite and pumice substituted cement mortars with artificial intelligence based applications

Artificial intelligence applications such as artificial neural networks (ANN) and adaptive network-based fuzzy inference system (ANFIS) are expressed as reliable methods that are widely used by many researchers to predict various properties in applications in the cement and concrete industry. For this reason, in this study, prediction models with different properties were developed by using ANN and ANFIS to predict the compressive strength of cement mortars. For these models, compressive strength results of seven different mortar samples, in which Portland cement and pumice and/or diatomite were substituted, were used on the 2nd, 7th, 28th and 90th hydration days. 168 data for the training and 28 data for the testing of the ANN and ANFIS models were used. In ANN and ANFIS models; 5 input parameters, namely hydration age, Portland cement, pumice, diatomite, water, and 1 output parameter, including compressive strength of mortars, were determined. Three different statistical methods preferred in the literature such as mean absolute percentage error (MAPE), root mean square error (RMSE) and coefficient of determination (R2) were used to compare the results. In addition, each test result was examined one by one and the differences and % changes between the estimated compressive strengths and the actual values on the 2nd, 7th, 28th and 90th hydration days were determined in order to see the errors more clearly. According to the data obtained, it was determined that the Elman backpropagation neural network model among the ANN models and the Psig membership function model among the ANFIS models were the best model. In addition, it is thought that the models are powerful and useful, therefore, the compressive strength estimations of cement mortars prepared with pumice and diatomite substitution with both ANN and ANFIS models can be performed quite well in a very small error and in a short time.