FLOW CHARACTERISTICS OF HYDRAULIC JUMP IN HORIZONTAL AND SLOPING CHANNELS USING ARTIFICIAL NEURAL NETWORK

摘要

The phenomenon of hydraulic jump in horizontal and sloping channel is so complex that despite considerable laboratory and prototype studies, estimation of its main characteristics in a generalized form is still difficult. The artificial neural network (ANN) approach would limit the need for costly and time consuming experiments. In the current study, many ANN models using multi-layer perceptron, invoking a back propagation algorithm (MLP/BP) and radial basis function using competitive learning (RBF/CL), were tried to predict the sequent depth, horizontal jump length, and relative energy loss of jumps in horizontal and sloping positive and adverse channels. Based on a pre-specified range of jump parameters, the input vector included upstream channel slope and inlet Froude number while the output vector included sequent depth, jump length and energy dissipation; were generated from the experimental data of different previous studies. A (2-6-3) MLP/BP model was concluded to be the optimal configuration amongst all other models regarding model performance. The predicted values of the model agree well with measurements and provided accurate prediction comparable to other empirical and/or theoretical equations. Sensitivity analysis was performed to investigate the importance of each input on the three output parameters.