Abstract
The management of coastal zones as a whole affects social and economic life and includes safeguards against extreme waves and floods. The accurate estimation of wave overtopping at coastal structures is therefore crucial to adequately protect people and infrastructure in these regions. This study employed artificial neural network-based (ANN) approaches with different algorithms, such as multilayer perceptron (MPNN), and general regression (GRNN), and support vector machine (SVM) for estimating the wave overtopping discharge at rubble mound structures featuring a straight slope. This study makes use of the new EurOtop database as its data source. Six distinct parameters (MSE, MAE, RMSE, SI, Ef and R) were utilized to assess the predictive performance of each model. Regarding the prediction of the wave overtopping discharge, the GRNNN produced exceptionally precise results. The SI of the GRNN was lower than that of the MPNN, and SVM by 600.11%, and 65.72%, respectively. In addition, the GRFNN model outperformed the other models in terms of efficiency. The Ef of the GRNN was higher than those of the MPNN, and SVM by 82.6%, and 3.0%, respectively.