Title: Prediction of discharge coefficient of combined weir-gate using ANN, ANFIS and SVM

Authors: Abbas Parsaie; Amir Hamzeh Haghiabi; Samad Emamgholizadeh; Hazi Mohammad Azamathulla

Addresses: Department of Water Engineering, Lorestan University, Khorramabad, Iran ' Water Engineering Department, Lorestan University, Khorramabad, Iran ' Department of Water and Soil Engineering, Shahrood University of Technology, Shahrood, Semnan Province, Iran ' Fiji National University, Suva, Fiji Islands

Abstract: Flow measurement is an important issue for developing the water conservation projects and evaluating the performance of irrigation and drainage networks. Weirs and gates are the most common structures which have been widely used for flow measurement. The main defects related them are deposition of suspended matter behind the weirs and accumulation of floating matter on water behind the gates, respectively. Therefore, the weir-gate structure has been proposed to solve them infirmities. In this study, predicting the discharge coefficient of weir-gate was considered using the artificial neural network (ANN), support vector machine (SVM) and adaptive neuro-fuzzy inference systems (ANFIS). For this purpose, the related dataset were collected from the literature. Assessing the performance of three models show that all of them have suitable accuracy, however, the SVM model with a coefficient of determination (R2 = 0.94) and root mean square of error (RMSE = 0.008) has the best performance in comparison with others. During the preparation of SVM it was found that the radial basic function as kernel function has best performance among the tested kernel functions. Sensitivity analysis of applied models showed that the ANN is the most sensitive model in comparison with others.

Keywords: flow measurement; hydraulic efficiency; irrigation and drainage networks; ANFIS.

DOI: 10.1504/IJHST.2019.102422

International Journal of Hydrology Science and Technology, 2019 Vol.9 No.4, pp.412 - 430

Received: 04 Nov 2017
Accepted: 11 Dec 2017

Published online: 18 Sep 2019 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article