Title :
Dynamic Behavior Modeling of Civil Structures Using Wavenets and Neural Networks: A Comparative Study
Author :
Perez-Ramirez, C.A. ; Amezquita-Sanchez, J.P. ; Valtierra-Rodriguez, M. ; Mejia-Barron, A. ; Dominguez-Gonzalez, A. ; Osornio-Rios, R.A. ; Romero-Troncoso, R.J.
Author_Institution :
HSPdigital-CA Mecatronica, Univ. Autonoma de Queretaro, San Juan del Rio, Mexico
Abstract :
Civil structures are known for having a non-linear and time-variant behavior, these features make a challenging task the use of linear methods for modeling the dynamical behavior since they only model time-invariant systems. To overcome this limitation, several approaches based on non-parametric methods have been proposed, however, the selection of the best-suited method for a particular case can be a complicated decision-making process. In this paper, a comparison between dynamic neural networks and wave nets for modeling the dynamic response of a five-bay space truss structure is presented, by using the structure response to a chirp signal, the models are created. Then, the root mean squared value (RMSE) is employed for determining the model that best approximates the dynamic behavior. An experimental study is carried out in order to validate the models efficiency and their accuracy.
Keywords :
approximation theory; condition monitoring; mean square error methods; neural nets; structural engineering computing; supports; RMSE; chirp signal; civil structures; decision-making process; dynamic behavior approximation; dynamic behavior modeling; dynamic neural networks; five-bay space truss structure; nonlinear time-variant behavior; nonparametric methods; root mean squared value; structure response; wavenets; Accuracy; Computational modeling; Data models; Mathematical model; Neurons; Predictive models; Training; Civil Structures; Dynamic Modeling; NARX; TDNN; Wavenets;
Conference_Titel :
Mechatronics, Electronics and Automotive Engineering (ICMEAE), 2014 International Conference on
DOI :
10.1109/ICMEAE.2014.33
