| Autor: |
Hieu NGUYEN-TRAN, Thanh BUI-TIEN, Magd ABDEL WAHAB, Dung BUI-NGOC |
| Jazyk: |
English<br />French |
| Rok vydání: |
2020 |
| Předmět: |
|
| Zdroj: |
Journal of Materials and Engineering Structures, Vol 7, Iss 4 (2020) |
| Druh dokumentu: |
article |
| ISSN: |
2170-127X |
| Popis: |
Structural Health Monitoring is a process of continuous evaluation of infrastructure status. In order to be able to detect the damage status, data collected from sensors have to be processed to identify the difference between the damaged and the undamaged states. In recent years, convolution neural network has been applied to detect the structural damage and with positive results. This paper proposes a new method of damage detection using combination of deep neural networks. The method uses a convolution neural network to extract deep features in time series and Long Short Term Memory network to find a statistically significant correlation of each lagged features in time series data. These two types of features are combined to increase discrimination ability compared to deep features only. Finally, the fully connected layer will be used to classify the time series into normal and damaged states. The accuracy of damaged detection was tested on a benchmark dataset from Los Alamos National Laboratory and the result shows that hybrid features provided a highly accurate damage identification. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
