چكيده به لاتين
Abstract:
The durability of concrete is far broader than its strength. Concrete durability is more complex because of different mechanisms and experiments. Concrete durability is assessed through durability tests, which usually require a lot of time and cost. One of the ways to reduce these costs is to use modeling in concrete durability. In this research, the performance of artificial neural network, ANFIS neuro-fuzzy model and Genetic Programing model for predicting the permeability of concrete against chloride ion has been investigated through rapid chloride permeability test (RCPT). For this modeling, 141 data obtained from other papers were used and for estimating the created models, two parameters of mean squared error and correlation coefficient were used. The results of the study and mean square error and correlation coefficient showed that both neural network model and ANFIS model have a very high ability to predict the permeability of concrete against chloride ion based on RCPT test results. For the artificial neural network model, the mean square error for the total data was 0.0033 and the correlation coefficient was 0.98 which indicates the proper performance of the neural network model. For the ANFIS model, the mean square error and correlation coefficient were 0.00049 and 0.99, which indicates that the ANFIS model has a better performance than the neural network model. For the Genetic Programing Model (GP), the mean squared error and correlation coefficient were 0.0168 and 0.82, respectively, indicating that this model has lower ability to predict the permeability of concrete against chloride ion than the other two models.
Keywords: Concrete durability, artificial neural network,ANFIS model, Genetic Programing, RCPT test
