Soil parameter identification from in situ measurement using a genetic algorithm and a principle component analysis

This paper is dedicated to the parameter identification of the Mohr-Coulomb constitutive model from horizontal displacements of a sheet pile wall. An optimization method based on a genetic algorithm is proposed. This robust optimization method localizes quickly a set of solutions close to the optimum values. Then, the identified set is described by a principal component analysis. This new inverse analysis method based on a genetic algorithm and a principle component analysis permits to define an ellipsoid set of solutions which is characteristic of the soil properties for geotechnical studies. © 2007 Taylor & Francis Group.