A pareto design of evolutionary hybrid optimization of ANFIS model in prediction abutment scour depth

In this paper, a novel pareto evolutionary structure of adaptive neuro-fuzzy inference system (ANFIS) network is presented for abutment scour depth predicting. The genetic algorithm (GA) and singular value decomposition (SVD) is utilized in optimizing design of nonlinear antecedent parts and linear consequent parts of TSK-type of fuzzy rules simultaneously in ANFIS design for the first time. To this end, first the parameters affecting the scour in the vicinity of abutments are detected. After that, 11 ANFIS-GA/SVD models are introduced through the combination of the parameters affecting the scour. Based on the modeling results, the ANFIS-GA/SVD models predict the scour around abutments with a reasonable accuracy. The superior model forecasts more than 63% of scours with an error of less than 8%. The correlation coefficient (R) for the model is computed roughly 0.978. The value of the average discrepancy ratio for the model is obtained 0.981. In addition, the results of the sensitivity analysis demonstrate that the Froude number (Fr) and the ratio of the flow depth to the radius of the scour hole (h/L) are the most noticeable parameters affecting the scour depth in the vicinity of the abutments. Ultimately, a comparison between the superior model and the previous studies are presented which reveal that the current study has better performance to predict scour depth around abutments.