摘要：Stilling basins are one of important structures for the control of hydraulic jumps. Determining the secondary depth and length of the jump is important to supply the correct hydraulic design. In this research, neural networks and numerical simulations were used to calculate the parameters for hydraulic jumps. A back-propagation neural network with two middle layers was designed, in which the input parameters included initial jump depth, flow rate, and a Froude number for the initial depth and width of the primary section moving to the secondary section. The output parameters were the length of the jump and the secondary depth. To perform threedimensional numerical simulations and solve the equation turbulence k-s, fluent software was used. Then, the result of the training network with the input of laboratory data was compared with the threedimensional (3-D) numerical simulation. This study demonstrated that the neural network led to better precision in approximating parameters. The MAPE of the neural network was 0.058%. The mean absolute percent error of the numerical simulation results was 2.41%. We used Topsis method to determine the best method and it showed that the ideal method to design is numerical model.
