Please use this identifier to cite or link to this item:
Title: Prediction of pore-water pressure using radial basis function neural network
Authors: Rahardjo, Harianto
Mustafa, M. R.
Rezaur, R. B.
Isa, M. H.
Issue Date: 2012
Source: Mustafa, M. R., Rezaur, R. B., Rahardjo, H., Isa, M. H. (2012). Prediction of pore-water pressure using radial basis function neural network. Engineering geology, 135-136, 40-47.
Series/Report no.: Engineering geology
Abstract: Knowledge of soil pore-water pressure variation due to climatic changes is fundamental for slope stability analysis and other problems associated with slope stability issues. This study is an application of Radial Basis Function Neural Network (RBFNN) modeling for prediction of soil pore-water pressure responses to rainfall. Time series data of rainfall and pore-water pressures were used to develop the RBFNN prediction model. The number of input neurons was decided by the analysis of auto-correlation between pore-water pressure data and cross-correlation between rainfall and pore-water pressure data. Establishing the number of hidden neurons by method of self learning network architecture determination and also by trial and error method was examined. A number of statistical measures were used for the evaluation of the network performance. Prediction results with a network architecture of 8–10–1 and a spread σ = 3.0 produced the lowest error measures (MSE, RMSE, MAE), highest coefficient of efficiency (CE) and coefficient of determination (R2). The results suggest that RBFNN is suitable for mapping the non-linear, complex behavior of pore-water pressure responses to rainfall. Guidelines for choosing the number of input neurons and eliminating possibility of model over-fitting are also discussed.
DOI: 10.1016/j.enggeo.2012.02.008
Rights: © 2012 Elsevier B.V.
Fulltext Permission: none
Fulltext Availability: No Fulltext
Appears in Collections:CEE Journal Articles

Citations 10

Updated on Mar 6, 2021

Citations 10

Updated on Mar 7, 2021

Page view(s) 50

Updated on Apr 16, 2021

Google ScholarTM




Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.