Efficient system reliability analysis of earth slopes based on support vector machine regression model

Metya, S., Mukhopadhyay, T., Adhikari, S. and Bhattacharya, G. (2017) Efficient system reliability analysis of earth slopes based on support vector machine regression model. In: Samui, P., Sekhar, S. and Balas, V. E. (eds.) Handbook of Neural Computation. Academic Press: Amsterdam, pp. 127-143. ISBN 9780128113189 (doi: 10.1016/B978-0-12-811318-9.00007-7)

Abstract

This chapter presents a surrogate-based approach for system reliability analysis of earth slopes considering random soil properties under the framework of limit equilibrium method of slices. The support vector machine regression (SVR) model is employed as a surrogate to approximate the limit-state function based on the Bishop's simplified method coupled with a nonlinear programming technique of optimization. The value of the minimum factor of safety and the location of the critical slip surface are treated as the output quantities of interest. Finally, Monte Carlo simulation in combination with Latin hypercube sampling is performed via the SVR model to estimate the system failure probability of slopes. Based on the detailed results, the performance of the SVR-based proposed procedure seems very promising in terms of accuracy and efficiency.

• Item Type: Book Sections
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Adhikari, Professor Sondipon
• Authors: Metya, S., Mukhopadhyay, T., Adhikari, S., and Bhattacharya, G.
• College/School: College of Science and Engineering > School of Engineering > Infrastructure and Environment
• Publisher: Academic Press
• ISBN: 9780128113189
• Published Online: 21 July 2017