Prediction of Backwater Level of Bridge Constriction using ANN

Abstract

Bridge constriction in channels usually causes afflux which results in increase in backwater level well above the normal level and may possibly result in overflow on the flood plain surrounding the channel during flooding period. This paper uses Artificial Neural Network to predict the afflux based on the parameters including coefficient of frictions of main channel (nmc) and of floodplain (nfp), bridge width (b) and flow discharge (Q). A Multi-Layer Perceptron (MLP) ANN is used to predict the afflux using these parameters. The training and testing data are the result of experimental investigation. It is observed that the afflux values predicted by the ANN model are very accurate compared to the experimentally measured values with a Normalized Mean Square Error (NMSE) of 0.002 and a Correlation Coefficient of 0.999. The developed ANN model can be used safely to conduct a parametric study to investigate the influence of the parameters nmc, nfp, b and Q on the afflux of a bridge constriction with piers.