Artificial neural networks approach for predicting the stability of cantilever RC retaining walls

Research output: Contribution to journalArticle

Abstract

The determination of the stability of cantilever reinforced concrete (RC) retaining walls is an important task in geotechnical engineering. This paper examines the potential of artificial neural network (ANN) for predicting the external stability of cantilever RC retaining walls. Two types of ANN model used as follows: ANN model (1 output) and ANN model (3 output). ANN model (1 output) involves a separate calculation for each output parameter while ANN model (3 output) involves calculations that take into account all the parameters simultaneously output. There are 235 different designs of cantilever RC retaining walls with three input parameters used for the prediction of safety factors such as height of the wall, angle of slope, and surcharge. The output parameters consist of the external stability namely: factors of safety (FOS) for sliding, overturning, and bearing capacity. Determination coefficient (R<sup>2</sup>) and root mean square error (RMSE) have been used for evaluation of prediction accuracy for both ANN models. The results of the study indicated that when compared in terms of R<sup>2</sup> value, ANN model (1 output) shows better performance than ANN model (3 output) does in predicting external stability because the value of R<sup>2</sup> is closer to one for all output parameters. Meanwhile, when compared in terms of RMSE value, ANN model (3 output) shows better performance than ANN model (1 output) since the value of RMSE for output parameters were closer to zero.

Original languageEnglish
Pages (from-to)26005-26014
Number of pages10
JournalInternational Journal of Applied Engineering Research
Volume10
Issue number10
Publication statusPublished - 2015

Fingerprint

Retaining walls
Reinforced concrete
Neural networks
Mean square error
Geotechnical engineering
Safety factor
Bearing capacity

Keywords

  • Artificial neural network (ANN)
  • Cantilever reinforced concrete (RC) retaining wall
  • Determination coefficient (R<sup>2</sup>)
  • External stability
  • Root mean square error (RMSE)

ASJC Scopus subject areas

  • Engineering(all)

Cite this

@article{47901a32b8c8457f92a388daac77924f,
title = "Artificial neural networks approach for predicting the stability of cantilever RC retaining walls",
abstract = "The determination of the stability of cantilever reinforced concrete (RC) retaining walls is an important task in geotechnical engineering. This paper examines the potential of artificial neural network (ANN) for predicting the external stability of cantilever RC retaining walls. Two types of ANN model used as follows: ANN model (1 output) and ANN model (3 output). ANN model (1 output) involves a separate calculation for each output parameter while ANN model (3 output) involves calculations that take into account all the parameters simultaneously output. There are 235 different designs of cantilever RC retaining walls with three input parameters used for the prediction of safety factors such as height of the wall, angle of slope, and surcharge. The output parameters consist of the external stability namely: factors of safety (FOS) for sliding, overturning, and bearing capacity. Determination coefficient (R2) and root mean square error (RMSE) have been used for evaluation of prediction accuracy for both ANN models. The results of the study indicated that when compared in terms of R2 value, ANN model (1 output) shows better performance than ANN model (3 output) does in predicting external stability because the value of R2 is closer to one for all output parameters. Meanwhile, when compared in terms of RMSE value, ANN model (3 output) shows better performance than ANN model (1 output) since the value of RMSE for output parameters were closer to zero.",
keywords = "Artificial neural network (ANN), Cantilever reinforced concrete (RC) retaining wall, Determination coefficient (R<sup>2</sup>), External stability, Root mean square error (RMSE)",
author = "Rohaya Alias and Anuar Kasa and Taha, {Mohd. Raihan}",
year = "2015",
language = "English",
volume = "10",
pages = "26005--26014",
journal = "International Journal of Applied Engineering Research",
issn = "0973-4562",
publisher = "Research India Publications",
number = "10",

}

TY - JOUR

T1 - Artificial neural networks approach for predicting the stability of cantilever RC retaining walls

AU - Alias, Rohaya

AU - Kasa, Anuar

AU - Taha, Mohd. Raihan

PY - 2015

Y1 - 2015

N2 - The determination of the stability of cantilever reinforced concrete (RC) retaining walls is an important task in geotechnical engineering. This paper examines the potential of artificial neural network (ANN) for predicting the external stability of cantilever RC retaining walls. Two types of ANN model used as follows: ANN model (1 output) and ANN model (3 output). ANN model (1 output) involves a separate calculation for each output parameter while ANN model (3 output) involves calculations that take into account all the parameters simultaneously output. There are 235 different designs of cantilever RC retaining walls with three input parameters used for the prediction of safety factors such as height of the wall, angle of slope, and surcharge. The output parameters consist of the external stability namely: factors of safety (FOS) for sliding, overturning, and bearing capacity. Determination coefficient (R2) and root mean square error (RMSE) have been used for evaluation of prediction accuracy for both ANN models. The results of the study indicated that when compared in terms of R2 value, ANN model (1 output) shows better performance than ANN model (3 output) does in predicting external stability because the value of R2 is closer to one for all output parameters. Meanwhile, when compared in terms of RMSE value, ANN model (3 output) shows better performance than ANN model (1 output) since the value of RMSE for output parameters were closer to zero.

AB - The determination of the stability of cantilever reinforced concrete (RC) retaining walls is an important task in geotechnical engineering. This paper examines the potential of artificial neural network (ANN) for predicting the external stability of cantilever RC retaining walls. Two types of ANN model used as follows: ANN model (1 output) and ANN model (3 output). ANN model (1 output) involves a separate calculation for each output parameter while ANN model (3 output) involves calculations that take into account all the parameters simultaneously output. There are 235 different designs of cantilever RC retaining walls with three input parameters used for the prediction of safety factors such as height of the wall, angle of slope, and surcharge. The output parameters consist of the external stability namely: factors of safety (FOS) for sliding, overturning, and bearing capacity. Determination coefficient (R2) and root mean square error (RMSE) have been used for evaluation of prediction accuracy for both ANN models. The results of the study indicated that when compared in terms of R2 value, ANN model (1 output) shows better performance than ANN model (3 output) does in predicting external stability because the value of R2 is closer to one for all output parameters. Meanwhile, when compared in terms of RMSE value, ANN model (3 output) shows better performance than ANN model (1 output) since the value of RMSE for output parameters were closer to zero.

KW - Artificial neural network (ANN)

KW - Cantilever reinforced concrete (RC) retaining wall

KW - Determination coefficient (R<sup>2</sup>)

KW - External stability

KW - Root mean square error (RMSE)

UR - http://www.scopus.com/inward/record.url?scp=84932084647&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84932084647&partnerID=8YFLogxK

M3 - Article

VL - 10

SP - 26005

EP - 26014

JO - International Journal of Applied Engineering Research

JF - International Journal of Applied Engineering Research

SN - 0973-4562

IS - 10

ER -