In the present study, intelligent methods including artificial neural network (ANN) and adaptive fuzzy neural inference system (ANFIS) have been investigated to evaluate the prediction of rockfill dam crest settlement. The accuracy of the methods used is compared with the central core based on crest settlement data obtained from 35 rockfill dams. Dam height and compressibility index were considered as input parameters. The compressibility index determines the general compression coefficient, which is determined by considering the compaction method of the substrate filling material and the quality of the foundation materials. The results of the present study showed that in the ANFIS model, the trampmf membership function is selected with two membership functions for each input with a value of C.C = 0.71, percentage, and MAE = 0.09%. Also, considering the results as a percentage, in the ANFIS model, the maximum amount of error is 34.64%, the minimum amount is 0.41% and the average is 12.01%. The best result in the neural network method will be obtained when 0.1 and 0.9 replace zero and one. The results showed that the slightest error occurs when using the Levenberh-Margaret post-publication law. To achieve the law of optimal education, other parameters affecting the neural network's performance have been kept constant, and by changing the rules of education, the network has been trained to repeat 1000 steps. For this purpose, a lattice with a hidden layer consisting of 7 nodes and a sigmoid transfer function was used. According to the results, it was observed that the error values in the neural network method are 1.88% in the minimum and 37.44% in the maximum, and also the average error was 14.23%.
