Abstract

This paper presents application of five soft-computing techniques, artificial neural networks, support vector regression, gene expression programming, grouping method of data handling (GMDH) neural network and adaptive-network-based fuzzy inference system, to predict maximum scour hole depth downstream of a sluice gate. The input parameters affecting the scour depth are the sediment size and its gradation, apron length, sluice gate opening, jet Froude number and the tail water depth. Six non-dimensional parameters were achieved to define a functional relationship between the input and output variables. Published data were used from the experimental researches. The results of soft-computing techniques were compared with empirical and regression based equations. The results obtained from the soft-computing techniques are superior to those of empirical and regression based equations. Comparison of soft-computing techniques showed that accuracy of the ANN model is higher than other models (RMSE = 0.869). A new GEP based equation was proposed.