Intelligent Modeling of Discharge Coefficient of Lateral Intakes

Intake structures have been widely used for flow diversion in the irrigation and drainage networks. In this paper, the multivariate adaptive regression splines (MARS), artificial neural network (ANN), and support vector machine (SVM) techniques were utilized for prediction of discharge coefficient (Cd) of lateral intakes. The experimental data pertaining to dimensionless parameters on Cd were collected to develop the models. The results indicated that the best performance in modeling is related to the MARS model with R2=0.98 and RMSE=0.023 and the MARS model outperforms the ANN and SVM models. The tangent sigmoid and radial basic functions were found to be the most efficient transfer and kernel functions for ANN and SVM, respectively. Moreover, Froude number (Fr1) and the ratio of the weir height to the upstream flow depth (P/d1) were the most effective factors for predicting Cd. Evaluation of the performance of applied models in term of developed discrepancy ratio (DDR) index shows that the minimum data dispersivity is related to the MARS model