The Library
Deep learning for structural health monitoring
Tools
Tools
Tools
Pamuncak, Arya Panji (2021) Deep learning for structural health monitoring. PhD thesis, University of Warwick.
|
PDF
WRAP_Theses_Pamuncak_2021.pdf - Submitted Version - Requires a PDF viewer. Download (13Mb) | Preview |
Official URL: http://webcat.warwick.ac.uk/record=b3763676
Abstract
In recent years, Structural Health Monitoring (SHM) has attracted significant attention due to its potential in providing effective maintenance strategy for infrastructure. However, interpreting the SHM data remains a challenging task. Model-based interpretation which utilises a behaviour model in the interpretation, requires experts in both developing the model and understanding the change in the model. On the other hand, data-based/model-free interpretation methods reduce the complexity since no physical model is utilised. However, expertise is required in performing data-based interpretation methods due to the need of feature extraction. This thesis is motivated to develop a deep learning-based data interpretation method that can learn features automatically, thereby minimising the required expertise.
In this thesis, a deep learning-based method for estimating load capacity of bridges from bridges’ images is developed. Data labelling is performed using information from National Bridge Inventory (NBI) database. Parametric study is performed to further investigate the method.
A deep learning-based method that utilises correlation between two or more sensor measurements is proposed. This method employs raw measurement data from sensors. The proposed method is implemented for estimating structural responses by using measurements from other sensor as the input. The proposed method is compared with other machine learning methods and the method outperforms the other methods.
Two damage detection approaches utilising deep learning techniques are discussed: novelty detection and multiclass classification. Both frameworks successfully predict the presence of the damage that could not be detected by a frequency-based method.
An approach that combines deep learning with Moving Principal Component Analysis (MPCA) as an existing damage detection method is introduced. Experimental data collected from a laboratory-scale bridge are employed as a case study to validate the method. A series of investigation on parameters used in both MPCA and deep learning architecture are conducted in order to observe the method.
| Item Type: | Thesis (PhD) | ||||
|---|---|---|---|---|---|
| Subjects: | Q Science > Q Science (General) Q Science > QA Mathematics > QA76 Electronic computers. Computer science. Computer software T Technology > TA Engineering (General). Civil engineering (General) |
||||
| Library of Congress Subject Headings (LCSH): | Structural health monitoring, Deep learning (Machine learning), Neural networks (Computer science), Structural analysis (Engineering) | ||||
| Official Date: | June 2021 | ||||
| Dates: |
|
||||
| Institution: | University of Warwick | ||||
| Theses Department: | School of Engineering | ||||
| Thesis Type: | PhD | ||||
| Publication Status: | Unpublished | ||||
| Supervisor(s)/Advisor: | Laory, Irwanda ; Guyo, Weisi | ||||
| Sponsors: | Lembaga Pengelola Dana Pendidikan | ||||
| Format of File: | |||||
| Extent: | xiv, 224 leaves : illustrations | ||||
| Language: | eng |
Request changes or add full text files to a record
Repository staff actions (login required)
 |
View Item |
Downloads
Downloads per month over past year
