Utilization of a least square support vector machine (LSSVM) for slope stability analysis

TLDR

Least square support vector machine (LSSVM) is grounded in statistical learning theory and employs regression and classification techniques. This study investigates the capability of LSSVM for slope stability analysis. The authors model the factor of safety as a regression problem and the stability status as a classification problem using inputs such as unit weight, cohesion, friction angle, slope angle, height, and pore water pressure coefficient, develop equations, and compare the model with an artificial neural network. The LSSVM produces probabilistic outputs and is shown to be a robust model for slope stability analysis.

Abstract

This paper examines the capability of a least square support vector machine (LSSVM) model for slope stability analysis. LSSVM is firmly based on the theory of statistical learning, using regression and classification techniques. The Factor of Safety (FS) of the slope has been modelled as a regression problem, whereas the stability status (s) of the slope has been modelled as a classification problem. Input parameters of LSSSVM are: unit weight (γ), cohesion (c), angle of internal friction (ϕ), slope angle (β), height (H) and pore water pressure coefficient (ru). The developed LSSVM also gives a probabilistic output. Equations have also been developed for the slope stability analysis. A comparative study has been carried out between the developed LSSVM and an artificial neural network (ANN). This study shows that the developed LSSVM is a robust model for slope stability analysis.