In this paper, an approach for constitutive material modelling is presented, which is based on a combination of three soft computing strategies (uncertainty, artificial neural networks, and swarm intelligence). The uncertainty model fuzziness is applied to take imprecise measurements and imprecise boundary conditions of experimental investigations into account. Data series obtained from experimental tests are described as fuzzy processes, containing interval processes and deterministic processes as special cases. Feed forward and recurrent neural network approaches are presented, which can be used for three-dimensional constitutive material models in finite element analyses. For direction dependent material behaviour, special network structures are developed. Particle swarm optimization (PSO) is applied for identification of the network parameters. A PSO approach for deterministic parameters is extended to handle fuzzy network parameters. Direct and indirect network training strategies are presented using uncertain stress-strain or load-displacement data, respectively. For network training and validation, a concept for computational steering of experimental tests is introduced.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £42 +P&P)