Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
36331.pdf (2.28 MB)
ETD Abstract Container
Abstract Header
A Method of Structural Health Monitoring for Unpredicted Combinations of Damage
Author Info
Butler, Martin A
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1575967420002943
Abstract Details
Year and Degree
2019, PhD, University of Cincinnati, Engineering and Applied Science: Civil Engineering.
Abstract
The current state of the art in structural health monitoring (SHM) is to compare a set of sensor values to previously established values, using some method to determine the most similar to positively identify the state. Some systems will merely establish that a change in the structure has occurred without positively identifying the system state. In order for most systems to identify a state, it must have been considered prior to the state (Helmicki et al. 2012). Those that do not require this previously determined state require intense computations following the detection of an abnormality, the solution space for damages is always large (He and Hwang, 2007). If multiple damage states are introduced, considering all possible combinations of them swiftly becomes untenable. This research examines the use of subdivided attractor artificial neural networks (SA-ANN) as a method for determining multiple damage states not considered prior to sensing. These networks receive signals from within themselves and from the other subnetworks, the subnetworks then are able to stabilize or destabilize each other depending on whether the physical states they represent are consistent with one another. This results in a new system that considers the state of a structure holistically, but is still able to discern multiple concurrent damage states. Test problems of two bridges are considered, a small pony truss bridge and a large cable stayed bridge. These bridges were divided into subsystems, and each individual damage state was modeled using SAP2000. Genetic algorithms (GA) were used to select strains to identify the individual damage states of the subsystems, and feedforward artificial neural networks (FF-ANN) were trained to identify damage to the subsystems based on these strains. These FF-ANN initialize the SA-ANN, which then converges to a state describing the physical state of the structure. Damage to multiple subsystems was also modeled; these were used to test the ability of the SHM system to identify damage states that were not known prior to the systems creation. The performance of the SHM system indicated that the pony truss bridge is a poor choice for implementation of the proposed system, the dearth of alternative load paths and tight correlation of the physical subsystem states usually result in failure to identify the less grievous damage state. Additionally, combinations of damage states that both resulted in a change in load path frequently would result in the collapse of the structure. The cable stayed example worked much better; the indeterminate structure endured unique alternative load paths for all damage states, and though the subsystems were connected by continuous girders and a deck, they were distinct enough that damage to one subsystem affected, but did not dominate, the neighboring subsystems. For all combinations of damaged subsystems tested the SHM system correctly identifies most of the states, with those subsystems with the most connections to other systems correctly identifying 95% of the test states.
Committee
James Swanson, Ph.D. (Committee Chair)
Thomas Burns, Ph.D. (Committee Member)
Ali Minai, Ph.D. (Committee Member)
Gian Andrea Rassati, Ph.D. (Committee Member)
Pages
110 p.
Subject Headings
Civil Engineering
Keywords
Structural Health Monitoring
;
Subdivided Attractor Networks
;
Artificial Neural Networks
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Butler, M. A. (2019).
A Method of Structural Health Monitoring for Unpredicted Combinations of Damage
[Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1575967420002943
APA Style (7th edition)
Butler, Martin.
A Method of Structural Health Monitoring for Unpredicted Combinations of Damage.
2019. University of Cincinnati, Doctoral dissertation.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1575967420002943.
MLA Style (8th edition)
Butler, Martin. "A Method of Structural Health Monitoring for Unpredicted Combinations of Damage." Doctoral dissertation, University of Cincinnati, 2019. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1575967420002943
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
ucin1575967420002943
Download Count:
207
Copyright Info
© 2019, some rights reserved.
A Method of Structural Health Monitoring for Unpredicted Combinations of Damage by Martin A Butler is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by University of Cincinnati and OhioLINK.