Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization
The academic community has paid much attention to the research topic of structural health monitoring (SHM) and damage identification for several decades, which flourishes the development of diverse damage identification approaches. The sensitivity-based damage identification enjoys popularity due to...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-01-01
|
| Series: | Advances in Civil Engineering |
| Online Access: | http://dx.doi.org/10.1155/adce/3961654 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832542585987530752 |
|---|---|
| author | Huihui Chen Xiaojing Yuan |
| author_facet | Huihui Chen Xiaojing Yuan |
| author_sort | Huihui Chen |
| collection | DOAJ |
| description | The academic community has paid much attention to the research topic of structural health monitoring (SHM) and damage identification for several decades, which flourishes the development of diverse damage identification approaches. The sensitivity-based damage identification enjoys popularity due to its clear physical meaning and long development history. However, this type of method faces challenges of fewer applications on the large-scale structure, slow convergence rate, and poor performance based on the sensitivity of the modal parameters. Aiming at the existing obstacles, this study enables to propose a novel method based on time series analysis model and improved sparse regularization technique for damage identification of the large-scale structure. Firstly, the model reduction technique is introduced to condense the unconcerned degrees of freedom (DOFs) of the finite element model (FEM) of the large-scale structure, and then the sensitivity of the autoregression (AR) coefficient with respect to the damage coefficient for the large-scale structure has been deduced, which can reduce the uncertainty, which comes from modal identification, in modal parameter sensitivity. Furthermore, the mean-value normalization strategy is incorporated into the sparse regularization solving process to improve the computational efficiency. The proposed method has been validated based on a numerical continuous rigid frame bridge and an experimental steel truss bridge. Compared to the moth-flame optimization (MFO) algorithm and traditional regularization methods, for both noise-free and noise polluted data, the iteration curves illustrate that the proposed method can achieve convergence within about 200 iterations, while the MFO algorithm is always trapped into local optima; meanwhile, the traditional regularization method needs more iterations or even cannot meet the preset tolerance. Furthermore, regarding the numerical example, the damage identification results show that the max identification errors of the proposed method are 6% among the three cases under noise-free and noise-polluted data, while the max identification errors for MFO and the regularization method are 15% and 10%, respectively. And for the experimental example, based on the limited sensors, the proposed method can accurately detect the damage location and quantify damage severity with a good accuracy, which means this method has good potential for actual engineering structure. To summarize, the findings highlight that the AR coefficient sensitivity combined with improve sparse regularization can effectively detect damage of large-scale structure, providing a feasible way to expand the application of the sensitivity-based method. |
| format | Article |
| id | doaj-art-72c213869d0048aebfe6ccb6cea48ebc |
| institution | Kabale University |
| issn | 1687-8094 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Wiley |
| record_format | Article |
| series | Advances in Civil Engineering |
| spelling | doaj-art-72c213869d0048aebfe6ccb6cea48ebc2025-02-04T00:00:03ZengWileyAdvances in Civil Engineering1687-80942025-01-01202510.1155/adce/3961654Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse RegularizationHuihui Chen0Xiaojing Yuan1Architectural Engineering InstituteKey Laboratory of Structural Engineering of Jiangsu ProvinceThe academic community has paid much attention to the research topic of structural health monitoring (SHM) and damage identification for several decades, which flourishes the development of diverse damage identification approaches. The sensitivity-based damage identification enjoys popularity due to its clear physical meaning and long development history. However, this type of method faces challenges of fewer applications on the large-scale structure, slow convergence rate, and poor performance based on the sensitivity of the modal parameters. Aiming at the existing obstacles, this study enables to propose a novel method based on time series analysis model and improved sparse regularization technique for damage identification of the large-scale structure. Firstly, the model reduction technique is introduced to condense the unconcerned degrees of freedom (DOFs) of the finite element model (FEM) of the large-scale structure, and then the sensitivity of the autoregression (AR) coefficient with respect to the damage coefficient for the large-scale structure has been deduced, which can reduce the uncertainty, which comes from modal identification, in modal parameter sensitivity. Furthermore, the mean-value normalization strategy is incorporated into the sparse regularization solving process to improve the computational efficiency. The proposed method has been validated based on a numerical continuous rigid frame bridge and an experimental steel truss bridge. Compared to the moth-flame optimization (MFO) algorithm and traditional regularization methods, for both noise-free and noise polluted data, the iteration curves illustrate that the proposed method can achieve convergence within about 200 iterations, while the MFO algorithm is always trapped into local optima; meanwhile, the traditional regularization method needs more iterations or even cannot meet the preset tolerance. Furthermore, regarding the numerical example, the damage identification results show that the max identification errors of the proposed method are 6% among the three cases under noise-free and noise-polluted data, while the max identification errors for MFO and the regularization method are 15% and 10%, respectively. And for the experimental example, based on the limited sensors, the proposed method can accurately detect the damage location and quantify damage severity with a good accuracy, which means this method has good potential for actual engineering structure. To summarize, the findings highlight that the AR coefficient sensitivity combined with improve sparse regularization can effectively detect damage of large-scale structure, providing a feasible way to expand the application of the sensitivity-based method.http://dx.doi.org/10.1155/adce/3961654 |
| spellingShingle | Huihui Chen Xiaojing Yuan Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization Advances in Civil Engineering |
| title | Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization |
| title_full | Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization |
| title_fullStr | Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization |
| title_full_unstemmed | Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization |
| title_short | Damage Identification in Large-Scale Structures Using Time Series Analysis and Improved Sparse Regularization |
| title_sort | damage identification in large scale structures using time series analysis and improved sparse regularization |
| url | http://dx.doi.org/10.1155/adce/3961654 |
| work_keys_str_mv | AT huihuichen damageidentificationinlargescalestructuresusingtimeseriesanalysisandimprovedsparseregularization AT xiaojingyuan damageidentificationinlargescalestructuresusingtimeseriesanalysisandimprovedsparseregularization |