Deep Neural Network-Based Storey Drift Modelling of Precast Concrete Structures Using RStudio
In this research, storey drift has been determined using Deep neural networks (DNN Keras). DNN Keras has various hyper tuning parameters (hidden layer, drop out layer, epochs, batch size and activation function) that make it capable to model complex problems. Building height, number of bays, number...
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
Pouyan Press
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/094e422d64534310bdb359e84b8e86c6 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | In this research, storey drift has been determined using Deep neural networks (DNN Keras). DNN Keras has various hyper tuning parameters (hidden layer, drop out layer, epochs, batch size and activation function) that make it capable to model complex problems. Building height, number of bays, number of storeys, time period, storey displacement, and storey acceleration were the input parameters while storey drift was the output parameter. The dataset consists of 288 models, out of 197 were used as training data and the remaining 91 were used as test data. 0.9598 correlation coefficient was observed for DNN Keras as compared to 0.8905 by resilient back-propagation neural networks (BPNN), indicating that DNN Keras has about 8 per cent improved efficiency in predicting storey drift. Wilcoxon signed-rank test (non-parametric test) was used to compare and validate the performance of DNN Keras and resilient BPNN algorithms. The positive results of this study point to the need for further research into the use of DNN Keras in structural and civil engineering. |
|---|