Wave transmission at low-crested structures using neural networks

Abstract

The European Union funded project DELOS was focused on wave transmission and an extensive database on low-crested rubble mound structures was generated. During DELOS, new empirical wave transmission formulae were derived. These formulae still showed a considerable scatter due to a limited number of parameters included. Neural networks based on a homogeneous database have resulted in a useful prediction model for wave overtopping within the EU project CLASH. The successful methodology of CLASH is applied within this study. The aim of this study is to improve the prediction of wave transmission in comparison to the empirical DELOS formulae with help of a prediction model based on neural networks. This paper gives a overview of the contents of the composed homogeneous database and gives insight in the capacity and accuracy of the final prediction model. The final prediction model includes 9 input parameters, which is more than at present in the existing hand-derived empirical formulae. The prediction model is accurate in predicting wave transmission for both smooth and mound low-crested structures.