Automation of calibration process adopting metaheuristic optimization method

Abstract

Optimization procedures offer a possibility for time-effective determination of input parameters values for complex soil constitutive models. The following paper presents a combination of the metaheuristic Particle swarm optimization method (PSO) and commercially available solver based on the finite element method (FEM). After the brief theoretical description, different alternatives to the PSO method are reviewed and tested. An optimal alternative is chosen and further used. In the second part of the paper, the combination PSO – FEM is utilized for a fully automatic derivation of input parameters values for the Hardening small strain model from pressuremeter tests. Predicted pressure-volume curves from the axisymmetric FE model gradually converge towards the measured curve until the accuracy criterion is reached.