Bridge backwater estimation: A comparison between artificial intelligence models and explicit equations

Estimation of bridge backwater has been one of the practical challenges of hydraulic engineering for decades. In this study, Genetic Programming (GP) was employed to estimate bridge backwater for the first time based on the conducted literature review. Furthermore, two new explicit equations were developed to predict bridge afflux using Genetic Algorithm (GA) and hybrid MHBMO-GRG algorithm. The performance of these models was compared with that of the Artificial Neural Network (ANN) and several explicit equations available in the literature considering both laboratory and field data. According to five considered performance evaluation criteria, two new explicit equations outperformed those available in the literature. Furthermore, GP and ANN achieved the best results with respect to four out of five considered criteria for training and testing datasets, respectively. To be more specific, ANN improved the Mean Square Error (MSE) and R-2 values of the explicit equation developed using GA by 44% and 12% for the testing data while GP enhanced the corresponding values by 62% and 9% for the training data. Finally, the results indicated that not only the artificial intelligence models considerably improved bridge afflux estimations in comparison to the explicit equations but also the suggested equations could significantly improve the accuracy of the available explicit equations. (C) 2021 Sharif University of Technology. All rights reserved.