This paper focuses on new methods that have been implemented and tested in the artificial neural network (ANN) recently developed by the authors to predict the wave overtopping discharge. An improved representation of the non-significant wave overtopping conditions can be obtained by extending the training database of the existing ANN to include all the non-zero values. The modification of the ANN architecture by the set-up of a classifier-quantifier scheme to improve the prediction of both low and high values of the overtopping is also investigated and discussed. The accuracy of the improved ANN is finally verified through the prediction of new datasets.
An Advanced and Improved Artificial Neural Network for the Prediction of Wave Overtopping / Zanuttigh B.; Formentin S.M.; Van Der Meer J.W.. - ELETTRONICO. - (2015), pp. 719-730. (Intervento presentato al convegno Coastal Structures and Solutions to Coastal Disasters Joint Conference 2015 tenutosi a Boston, USA nel 2015) [10.1061/9780784480304.076].
An Advanced and Improved Artificial Neural Network for the Prediction of Wave Overtopping
Zanuttigh B.;Formentin S. M.;
2015
Abstract
This paper focuses on new methods that have been implemented and tested in the artificial neural network (ANN) recently developed by the authors to predict the wave overtopping discharge. An improved representation of the non-significant wave overtopping conditions can be obtained by extending the training database of the existing ANN to include all the non-zero values. The modification of the ANN architecture by the set-up of a classifier-quantifier scheme to improve the prediction of both low and high values of the overtopping is also investigated and discussed. The accuracy of the improved ANN is finally verified through the prediction of new datasets.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
