Prediction of wave transmission coefficient using neural networks and experimental measurements

Abstract

In the present paper, an application of composite modeling is presented, for the estimation of wave transmission through flushing culverts. Specifically, experimental measurements are used and an Artificial Neural Network (ANN) is structured for the prediction of the transmission coefficient. Measurements were obtained from physical model tests that have been conducted in 2D and 3D experimental facilities, at the Laboratory of Harbour Works, at National Technical University of Athens. The derived ANN, despite the limited available experimental data, showed satisfactory behaviour and may be used as an alternative tool for a first estimation of the transmission coefficient through flushing culverts in harbour design in combination with proposed empirical equations or numerical models. © 2015 Taylor and Francis Group.