Calibration of the parameters for a hardening–softening constitutive model using genetic algorithms

The present paper introduces a genetic algorithm-based optimization technique to calibrate a nonlinear strain hardening–softening constitutive model for soils using five material parameters. The efficiency of the proposed technique is analyzed through the use of different GA techniques. The effects of elitism, crossover, and mutation, as well as population size, on the performance of the conventional GAs for this problem are investigated. Micro-genetic algorithms (mGAs) are chosen and tested for different population sizes. The mGAs with a population size of five yields the optimal parameter values after fewer function evaluations and capture the overall simulated or experimental behavior at every point in stress–strain and strain paths in triaxial compression. The proposed calibration technique is validated through comparison with the traditional calibration technique.