The effect of predicting discharge coefficient by neural network on increasing the numerical modeling accuracy of flow over side weir

Prediction and modeling of hydraulic phenomenon is an important part of hydraulic engineering activities. One of the applications of prediction and modeling is estimating the discharge coefficient for hydraulic structures. Side weirs are widely used for allocating and removing excess flow in water engineering projects. The governing equation on side weir hydraulic characteristics is spatially varied flow (SVF). Computer modeling of hydraulic characteristics this structure includes the calculation of water surface profiles and estimating the discharge coefficient ( sw Cd ). The numerical method was used to calculate the water surface profile and there are several ways to estimate the sw Cd , such as experimental formulas and computational intelligence techniques. In this paper, the Fourth Runge Kutta method is used for numerical solution of SVF and firstly to estimate sw Cd , some famous empirical equations are assessed. Among the empirical formulas, the Borghei is the most accurate one. To increase the accuracy of computer modeling, the Multilayer Neural network (MLP) is developed to estimate sw Cd . The result shows that using neural network to estimate sw Cd increases the accuracy of the final model about 16 %.

Keywords:

Multilayer Neural Network , Discharge Coefficient , Runge–Kutta Method , Side Weir