Servicios Personalizados
Revista
Articulo
Indicadores
- Citado por SciELO
Links relacionados
- Similares en SciELO
Compartir
Tecnia
versión impresa ISSN 0375-7765versión On-line ISSN 2309-0413
Resumen
ZAVALA, Carlos; DIAZ, Miguel y HONMA, Claudia. In the challenge of earthquake response prediction by neural network approach. Tecnia [online]. 2023, vol.33, n.1, pp.62-68. Epub 10-Jun-2023. ISSN 0375-7765. http://dx.doi.org/10.21754/tecnia.v33i1.1434.
Since the decade of the 1990 neural networks algorithms have been used for compute approximate solutions for different problems in engineering.
In the building behavior against loads is important to know its response. The behavior during earthquakes and the estimation of the response is quite difficult to compute due to the nonlinearity of geometry and material properties. Neural networks approach is a powerful tool for computing the response of structures with an appropriate learning process from big data of structural components. Even if some material parameters are unknown, learning on a neural network will be possible and will provide an estimation using collect information from experience and learning.
To make a learning process in this paper, we present a simple algorithm of back propagation implemented in python programming language where the output shows the decrease of the error and how the response start to learn from the beginning until the end of the process. The results show good agreement between the learning data set and predicted response after the neural network learning.
Palabras clave : confined masonry; neural networks; building response; dynamic earthquake response..