Automated Generation of Material Laws at Large Strains Using a Genetic Algorithm

Abstract

This work presents an approach for automatically generating material laws at large strain. To this end, a concept of material modelling based on directly connected rheological elements is applied. This concept allows the implementation of any material law defined by a rheological model by recursive evaluation of connection relations. Therewith a tree encoding for the rheological model is applied. The search for suitable material laws fitting experimental data is realized by a genetic algorithm, which generates rheological models and evaluates their quality with respect to the given measurements. The central concepts and the genetic operators are presented in this contribution. Finally, the method is verified with synthetic data and utilized to real experimental data.

Full document