Please use this identifier to cite or link to this item:
Title: Wave parameter estimation using neural networks
Authors: AGRAWAL, JD
Keywords: Neural Networks
Parameter Estimation
Regression Analysis
Water Waves
Issue Date: 2004
Publisher: Elsevier
Citation: Marine Structures 17(7), 536-550
Abstract: At different stages of wave analysis and wave-based structural design it becomes necessary to know interrelationships among certain characteristic wave parameters. This knowledge is currently obtained by using empirical spectral or statistical regression schemes, which may suffer from uncertainties and approximations. This paper presents an alternative to them, which is based on the neural network approach. Networks were developed in order to estimate values of average zero-cross wave period, peak-spectral period, maximum spectral energy density and maximum wave height from the given value of significant wave height and also to evaluate spectral width parameter from the spectral narrowness parameter. They were trained with respect to observed data at an offshore site along the east coast of India. The trained network when tested revealed that it formed a useful tool in exploring the interdependency in between the parameters. When the underlying relation structure was hazier and uncertain the neural networks were more accurate than the statistical non-linear regression methods in estimating one parameter from the specified value of the other.
ISSN: 0951-8339
Appears in Collections:Article

Files in This Item:
File Description SizeFormat 
5Deo4442.pdf275.92 kBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.