Prediction of Scour Depth Around Bridge Piers Using Evolutionary Neural Network

Abstract

Abstract An empirical formula based on evolutionary regression network is proposed in this paper for predicting the equilibrium depth of scour around bridge piers. The formula expresses the equilibrium scour depth as a function of variables including flow depth and mean velocity, critical flow velocity, median grain size and pier diameter. The empirical formula is developed by training and testing an evolutionary network using scour data available in the literature. The use of the evolutionary algorithm in developing the formula is informed by the need to reduce the model complexity while sacrificing its predictive accuracy. The results of performance comparisons with existing models showed that the proposed formula model produces reasonably accurate estimates of equilibrium scour depth with a much smaller number of fitting constants compared with backpropagation neural networks.