Adapted Genetic Algorithm Applied to Slope Reliability Analysis

Abstract

A new genetic algorithm (GA) was Presented with good convergence properties and a remarkable low computational load. Such features are achieved by dynamically tuning up the probabilities of crossover and mutation on the basis of the variability of the individuals' fitness values. This way, a brand new method to control and adjust the population diversity is obtained. The resulting GA attains quality solutions, offering an alternative to other global search techniques, as well as standard GAs. The new method is applied to solve the problem of reliability analysis in slope engineering, where the main aim is to obtain the lowest reliability index which corresponds to the most dangerous failure surface. For a given critical surface, an optimization model on slope reliability was deducted by Lagrange method, and computational load was reduced as well; the algorithm was adopted to search for the critical surface where a fitness function was suggested for slopes and genetic operators were controlled automatically so as to attain a good convergent but low complex property.