Application of Artificial Neural Network and M5 model tree for determination of discharge coefficients in broad crested weirs

Document Type : Research Article


Department of Water Engineering, Faculty of Agriculture, University of Tabriz, Tabriz, Iran


Broad crested weir is a simple structure for discharge measurement at water inlet channels. In addition, it is using as dam’s weir because of its geometrical shape and high weights. The entrance of some stepped weirs or chutes is designing in type of broad crested weir. Sometimes this structure is using as dam’s body. In this study, the ability of Artificial Neural Network (ANN) and M5 model tree in predicting of discharge coefficient (Cd) in broad crested weir have been considered and the results of these two models have been compared with nonlinear regression method. For this purpose four series of data resulting from different researches on rectangular broad crested weir have been used and the dimensionless parameters of H1/L and H1/P have been defined as entrance of ANN and M5 model tree and the target parameter (Cd) have been extracted as output of the models. Results showed that all the three method have a reasonable prediction for Cd (ANN: R=0.966 ، M5Rule: R=0.874 and Regression: R=0.84(but due to having simple equations with M5 Model tree, this method would be used as an efficient application for the estimation of discharge coefficient. Also, H1/L is the most important parameter in calculating the Discharge coefficient of Broad crested weir. M5 Model tree provides four linear equations (rules) for estimation of Cd. Non-linear regression analysis showed that point H1/L=0.22 is an intersection of the all curves for Cd variation.


Main Subjects