Prediction of pile scour in the ocean using Ann
Shailesh Namekar, A.R. Kambekar and M.C. Deo
Department of Civil Engineering, Indian Institute of Technology, Bombay, Powai, Mumbai 400 076.
The maximum depth of scour hole developed around a pile group in the ocean has been obtained using the artificial neural network approach. It requires input of four dimensionless groups of parameters, namely, Reynolds number, Keulegan-Carpenter number, Sediment number and Shield's parameter. Multi-Layered Perceptron as well as Radial Basis Function networks are considered. The testing results showed satisfactory performance of the developed network. The neural network approach yielded better results than the traditional statistical regression method indicating its superiority over the latter. The traditional Multi-Layered Perceptron produced more satisfactory results than the Radial Basis Function network.
Back
Application of Ann approach for prediction of tides
N. Vivekanandan
Dr. Mahalingam College of Engineering & Technology, Pollachi 642003
Prediction of tides is a pre-requisite for planning and design of ports and harbours. Tidal level prediction is usually done by harmonic analysis of observed data and finding the tidal constituents and it requires to be run for different seasons which are time consuming. In this context, Artificial Neural Network approach is used as an alternative method for prediction of tides at sub-ordinate stations. The paper presents the application of Back Propagation Network (BPN) and Cascade Correlation Network (CCN) for prediction of tides at Okha and Navalakhi stations of Gulf of Kutch. The paper presents that the results of predicted tides using BPN could be beneficially encapsulated in CCN so as to save enormous efforts involved in CPU time as well as long duration of boundary conditions at the reference stations to be prescribed. The methodology adopted in training and testing the BPN, CCN and the results obtained on the study are presented in this paper.
Back
Relating deep water waves with coastal waves using Ann
Ruchi Kalra¹, M. C. Deo¹, Raj Kumar² and Vijay K. Agarwal²
¹Department of Civil Engineering, Indian Institute of Technology, Bombay, Powai, Mumbai 400076.
²Oceanic Sciences Division, Meteorology and Oceanography Group, Space Applications Centre (ISRO), Ahmedabad, 380 015
A technique to obtain significant wave heights at coastal locations from their values sensed at deeper locations with the help of a satellite is discussed in this paper. This technique is based on the approach of neural networks. The applicability of the method was first checked by a highly satisfactory mapping between deep-water buoy measurements and corresponding shallow water buoy observations. The satellite data of significant wave height, average wave period and the wind speed were then given as input in order to obtain significant wave heights at a costal location along the west coast of India. This involved use of a common MLP network trained using variants of back-propagation algorithms. Qualitative as well as quantitative comparison of the network output with target observations showed usefulness of neural networks in such an application. Unlike satellite observations collection of buoy data is costly and hence it is generally resorted to fewer locations and for a smaller period of time. As shown in this study the network can be trained with samples of buoy data and can be further used for routine wave forecasting at coastal locations based on more permanent flow of satellite observations. The wave rider buoy can be discarded or redeployed elsewhere.
Back
On line tide prediction using artificial neural networks
A.M. Chitale¹, P. S. Pathak¹, V.S. Limaye² and S.N. Londhe²
¹Final year Students, Sinhgad College of Engineering, Pune
²Department of Civil Engineering, Sinhgad College of Engineering, Pune
The knowledge of tide prediction is very much essential in effective planning and implementation of projects related to coastal and harbor engineering problems. This paper presents an Artificial Neural Network approach for on line forecasting of the tidal levels atBrandywine station in Delaware City on East Coast of USA. Two separate models were developed, the first one for determination of water levels at 6 minutes time interval and the other one for prediction of high and low water levels. Three different algorithms namely, Conjugate Gradient Fletcher Reeves update (CGF), Broydan-Fletcher-Goldfarb-Shanno algorithm (BFG) and Levernberg-Marquardt (LM) algorithm were tried. The testing results were highly in agreement with the observed data as indicated by scatter plots and accompanying high values of correlation coefficients and two other error measures, namely, Coefficient of Efficiency (CE), Root Mean Squared Error (RMSE).
Back
Neural networks for estimation of ocean wave parameters
S. Mandal¹, Subba Rao² and D.H.Raju²
¹Ocean Engineering Division, National Institute of Oceanography, Dona Paula, Goa-403 004
²Dept of Applied Mechanics & Hydraulics, National Institute of Technology Karnataka, Surathkal, P.O.Srinivasnagar-575025
Ocean wave parameters play a significant role in the design of all coastal and offshore structures. In the present study, neural networks are used to estimate various ocean wave parameters from theoretical Pierson-Moskowitz spectra as well as measured ocean wave spectra off Mormugao, west coast of India. The correlations of neural network and Scott spectra are also compared. Once the network is trained, the ocean wave parameters can be estimated for unknown measured spectra, whereas significant wave height and spectral peak period are required to generate the Scott spectra. This study shows that the missing field wave data can be generated using trained network.
Back
Suspended sediment profiles derived from spectral attenuation coefficients measurements using neural network method
Geeta Varkey¹, T. Suresh², S.G. Prabhu Matondkar², Elgar Desa² and S.S. Kamath¹
¹National Institute of Technology, Surathkal, Karnataka
²National Institute of Oceanography, DonaPaula, Goa-403 004
Suspended sediment in water plays a major role in the operations of ports and harbor. Here we report a model to be used with an optical instrument that is primarily used for the measurements of inherent optical properties. The AC-9 (WetLabs, USA) is an instrument used for in-situ measurements of the inherent optical properties of the water, absorption and beam attenuation at different wavelengths. The aim is to establish a relationship between the measured profiles of optical properties and the measured total suspended matter values from water samples obtained at discrete depths at the same location. An artificial neural network (ANN) model has been used to derive suspended matter from the spectral values of beam attenuation coefficients measured using the AC-9 instrument. The ANN model has been trained using beam attenuation values obtained from the cruise in the Arabian Sea, SK-149 and the net was tested using the data from the cruise SK-152. The feed forward multi-layer perceptron neural network model was trained using the Levenberg-Marquardt algorithm and has provided us with encouraging results, which are in very close agreements with the measured values. This ANN model with eight input nodes, one hidden layer with five nodes and one output node can now be used to obtain profiles of suspended matter with the measurements of the AC-9 instrument, which will provide variations of concentrations at very fine depth intervals.
Back
|
