Artificial Bee Colony Algorithm Optimized Support Vector Regression for System Reliability Analysis of Slopes

Abstract

Probabilistic stability analysis is an effective way to take uncertainties into account in evaluating the stability of slopes. This paper presents an intelligent response surface method for system probabilistic stability evaluation of soil slopes. Artificial bee colony algorithm (ABC) optimized support vector regression (SVR) is used to establish the response surface to approximate the limit-state function. Then Monte Carlo simulation is performed via the ABC-SVR response surface to estimate system failure probability. The proposed methodology is verified in three case examples and is also compared with some well-known or recent algorithms for the problem. Results show that the new approach is promising in terms of accuracy and efficiency.