Use of a landslide problem for the identification parameters in geotechnics

Abstract

Soil behavior has been widely studied. However, methods for identifying parameters of soil behavior models have not been given equal importance. Identifying these parameters has posed a challenge for geotechnical calculations. Inverse analysis has emerged as an approach that aims to solve this problem by seeking to identify the optimal values of the desired parameters by minimizing the gap between the responses simulated by the model and in situ observations. The objective of this study has been to apply the principle of inverse analysis by exploiting the power of two optimization algorithms, the genetic algorithm and the hybrid genetic algorithm with tabu search method, in order to identify two parameters of the Mohr-Coulomb soil model, the cohesion and the friction angle. Both algorithms have been validated on a landslide case at the Boulakroud site in Skikda, Algeria. The results obtained have shown that the hybridization of the genetic algorithm with the tabu search method has proved to be more efficient, allowing a faster convergence towards the exact optimum of the problem. Conversely, the genetic algorithm alone has required a longer computation time to reach an optimum close to the exact optimum.