doi:10.3850/978-981-08-7920-4_S2-S71-cd

ABSTRACT

A new approach for identification and prediction of long-term structural behavior is presented. Dependencies between uncertain time-varying structural actions and responses can be formulated with artificial neural networks. Recurrent neural network sare developed to operate with uncertain data sequences. They are applied to identify stress-strain-time dependencies in uncertain data obtained from structural monitoring.This model-free material description enables touse recurrent neural networks for fuzzy data instead or asparts of material models with in the structural analysis based on the finite element method. The recurrent neural network based approach is verified by means of a numerical example. Additionally, an application is presented.

Keywords: Recurrent neural network, Fuzzy data, Model-free approach, Structural analysis, Uncertainty, Identification, Prediction, Textile reinforced concrete.

© 2011 Research Publishing Services. All rights reserved.