An evolutionary-based modelling approach for predicting the effective stress parameter in unsaturated soils

Ahangar Asr, A ORCID: https://orcid.org/0000-0002-8210-7519 and Javadi, A 2018, An evolutionary-based modelling approach for predicting the effective stress parameter in unsaturated soils , in: The 7th International Conference on Unsaturated Soils UNSAT2018, 3-5 August 2018, Hong Kong, China.

[img] PDF - Accepted Version
Restricted to Repository staff only

Download (484kB) | Request a copy

Abstract

Effective stress parameter affects the stress equation and is implemented to calculate the effective stress value which directly influences deformations and failures in unsaturated soils. Evolutionary polynomial regression is a data mining-based technique relying on evolutionary computing developed to search for structured and explicit polynomial models representing the behavior of complicated systems. The methodology uses a combination of genetic algorithm and the least square method to find best polynomial structures along with the corresponding parameters for each term in the structure. In this study an EPR model was developed using data from a set of unsaturated triaxial tests from literature. The developed model was used to predict the effective stress parameter for an unseen set of data to validate its generalization capabilities. The EPR model predictions were then compared to the experimental data revealing that the proposed model was capable of capturing and reproducing the underlying relations between the six considered input parameters and the effective stress parameter as the output, to a good level of accuracy.

Item Type: Conference or Workshop Item (Paper)
Schools: Schools > School of Computing, Science and Engineering > Salford Innovation Research Centre (SIRC)
Depositing User: Dr Alireza Ahangar Asr
Date Deposited: 04 Oct 2018 09:12
Last Modified: 04 Oct 2018 16:21
URI: http://usir.salford.ac.uk/id/eprint/48565

Actions (login required)

Edit record (repository staff only) Edit record (repository staff only)

Downloads

Downloads per month over past year