Parameter identification of a non-linear viscoelastic model for engineering polymers using a genetic algorithm

摘要

This paper presents an alternative approach for determining non-linear viscoelastic parameters of a constitutive model for glassy polymers using a multi-objective diversity control oriented genetic algorithm (MODCGA). The algorithm is a stochastic multi-objective optimisation technique. In this work, minimisation objectives are defined from errors between the experimental data (creep-recovery and constant strain rate compressive tests) and the simulation results. The parameters obtained by the MODCGA are found to be similar to that obtained previously using a knowledge-based iterative trial-and-error manual fitting method. However, the use of MODCGA requires much less human-computer interaction during the optimisation process and more refined solutions can be obtained without initial guess values being provided.