Bridge health condition assessment using instrumented moving vehicles
Loading...
Journal Title
Journal ISSN
Volume Title
School of Engineering |
Doctoral thesis (article-based)
| Defence date: 2023-10-06
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Author
Date
2023
Major/Subject
Mcode
Degree programme
Language
en
Pages
119 + app. 109
Series
Aalto University publication series DOCTORAL THESES, 144/2023
Abstract
Recently, there has been a growing interest in the indirect method of utilizing instrumented vehicles for bridge health monitoring. This approach only necessitates a few sensors mounted on vehicles instead of bridges, making it cost-effective, user-friendly, and easy to maintain. This thesis investigates the extraction of bridge frequencies and damage detection through numerical simulations and laboratory experiments. Firstly, to identify the bridge's frequencies from the vehicle's accelerations, a practical 3D vehicle model is developed in place of the frequently used quarter-car or half-car models. To filter out the vehicle's dynamic information, newly formulated equations are used to compute the vehicle's contact-point response. The influence of road roughness is eliminated by employing residual contact-point response of the vehicle's front and rear wheels. The effectiveness of the proposed method is confirmed by the results of numerical simulations considering various influencing factors. To improve the sensitivity of indirectly identified bridge frequencies to local damage, another parked truck is employed at different positions on the bridge for elaborate model updating and damage identification. Numerical simulation results indicate that the damage can be detected, localized, and quantified while several impacting factors are included. Secondly, using support vector machine models, this thesis shows that both low- and high-frequency responses from the instrumented vehicle contain the bridge's dynamic information and can therefore be employed for damage detection. Mel-frequency cepstral coefficients (MFCCs), originally derived from acoustic recognition, are extracted from the vehicle's accelerations as damage-sensitive features. The viability of MFCCs for damage detection has been confirmed through laboratory experiments with a model truck and a U-shaped beam. In addition, to address the challenge of obtaining damaged labels in practical engineering, this thesis introduces the assumption accuracy method. Instead of labeled data, it aims to determine the bridge's health states using accuracy values of binary classifications. Two laboratory beams and vehicles with two different weights have been used to validate the proposed method. Results show that when MFCCs are applied as a dimension reduction technique, the accuracy remains around 0.5 for a healthy bridge and approaches 1.0 once the bridge is damaged. Finally, in this thesis, bridge damage detection is accomplished in a real-time manner using the responses from an instrumented vehicle. This demonstrates that the passing vehicle's short-time vibrations can be employed to assess the bridge's health condition. By implementing the proposed damage indicator using the deep auto-encoder, the bridge's damage information can be automatically identified. A novel index called identified damage ratio (IDR) is proposed as an indicator for quantifying damage severity. Laboratory experiments show that as damage severity increases, the IDR initially rises significantly and then gradually approaches 100%.Description
Supervising professor
Lin, Weiwei, Associate Prof., Aalto University, Department of Civil Engineering, FinlandKeywords
structural health monitoring, vehicle-bridge interaction, indirect method, damage detection, machine learning
Other note
Parts
-
[Publication 1]: Li, Zhenkun; Lin, Weiwei; Zhang, Youqi. Bridge Frequency Scanning Using the Contact-Point Response of an Instrumented 3D Vehicle: Theory and Numerical Simulation. Structural Control and Health Monitoring, 2023, 3924349, June 2023.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-202308114759DOI: 10.1155/2023/3924349 View at publisher
-
[Publication 2]: Li, Zhenkun; Lan, Yifu; Lin, Weiwei. Indirect damage detection for bridges using sensing and temporarily parked vehicles. Engineering Structures, 291, 116459, September 2023.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-202306304343DOI: 10.1016/j.engstruct.2023.116459 View at publisher
-
[Publication 3]: Li, Zhenkun; Lin, Weiwei; Zhang, Youqi. Drive-by bridge damage detection using Mel-frequency cepstral coefficients and support vector machine. Structural Health Monitoring, 22(5), 3302-3319, February 2023.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-202302082055DOI: 10.1177/14759217221150932 View at publisher
-
[Publication 4]: Li, Zhenkun; Lan, Yifu; Lin, Weiwei. Investigation of Frequency-Domain Dimension Reduction for A2M-Based Bridge Damage Detection Using Accelerations of Moving Vehicles. Materials, 16(5), 1872, February 2023.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-202302282199DOI: 10.3390/ma16051872 View at publisher
-
[Publication 5]: Li, Zhenkun; Lin, Weiwei; Zhang, Youqi. Real-time drive-by bridge damage detection using deep auto-encoder. Structures, 47, 1167-1181, January 2023.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-202212146962DOI: 10.1016/j.istruc.2022.11.094 View at publisher