Aplicación de redes neuronales recurrentes para acelerar el análisis no lineal de estructuras.

Abstract

Nonlinear structural analysis is a performance-based engineering tool for the validation of proposed designs either for new structures or to evaluate existing structures. The application of this analysis for structural systemscomposed of a large number of degrees of freedom, such as high-rise buildings, bridges and even dams, the computational analysis can be time - consuming. Computational burden may result cumbersome for parametric studies. In order to reduce this computational cost, a practical methodology consisting in the application of neural networks and recurrent neural networks (RNN) is proposed in this work. Within the group of recurrent neural networks, long short-term memory (LSTM) is applied to accelerate the nonlinear analysis of structures. By creating a database for training and testing the neural network, it will be used not only to learn the behavior of the analysis, but also to make reliable predictions in addition to the main goal of reducing the computational burden. Furthermore, the recommendation to the applications and research of different types of neural networks for the solution of engineering problems.