Use of Artificial Neural Networks to predict strong ground motion duration of interplate and inslab mexican Earthquakes for soft and firm soils
Artificial neural network models are developed to predict strong ground motion duration of sub- duction events for soft and firm soils. To train the artificial neural network a database with a total of 3153 seismic records with two horizontal components for interplate and inslab earthquakes is emplo...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2022 |
| País: | México |
| Institución: | UNIVERSIDAD NACIONAL AUTÓNOMA DE MÉXICO |
| Repositorio: | Geofísica Internacional |
| Idioma: | español |
| OAI Identifier: | oai:revistagi.geofisica.unam.mx:article/1414 |
| Acceso en línea: | http://revistagi.geofisica.unam.mx/index.php/RGI/article/view/1414 |
| Access Level: | acceso abierto |
| Palabra clave: | Red neuronal artificial Duración del movimiento fuerte del terreno Expresiones empíricas y México Eventos de subducción Artificial neural network Strong ground motion duration Subduction events Empirical expressions and Mexico |
| Sumario: | Artificial neural network models are developed to predict strong ground motion duration of sub- duction events for soft and firm soils. To train the artificial neural network a database with a total of 3153 seismic records with two horizontal components for interplate and inslab earthquakes is employed. The principal component method is used to carry out a dimensionality reduction of the input parameters to develop the artificial neural network models. The predicted values of the strong ground motion duration trained by the artificial neural network models are compared with those estimated with empirical expressions. In general, the strong ground motion duration predicted with the artificial neural networks follows the same tendency of that calculated with the empirical equa- tions, although in some cases, the strong ground motion duration predicted by using the artificial neural network models presents sudden changes in its behavior. For this reason, it is recommended to carry out several verifications of the trained artificial neural network models before using them for further engineering applications, for example the simulation of synthetic records or the evaluation of seismic damage indices. |
|---|