TY - JOUR
T1 - Damage identification in timber bridges utilising the damage index method and neural network ensembles
AU - Dackermann, Ulrike
AU - Li, Jianchun
AU - Samali, Bijan
PY - 2009
Y1 - 2009
N2 - Many of Australia's timber bridges are in aged and decayed conditions. In order to ensure the reliability of these structures and the safety of the public, condition assessment, damage detection and safety evaluation is necessary. This paper presents a damage identification procedure, which is based on global change of vibration characteristics of a structure. The developed method utilises the damage index (DI) method in combination with neural network techniques to identify damage in numerical and experimental timber beam structures. The neural network ensemble approach is utilised in order to respect important diversities of different modes and to integrate individual characteristics of vibrational mode separated damage features. The method considers field testing issues associated with measurement noise, limited number of sensor arrays and environmental fluctuations. The results of damage detection using the proposed approach demonstrate its ability to determine the location and severity of all present damage cases. The outcomes show that the developed damage detection method is effective, robust and reliable.
AB - Many of Australia's timber bridges are in aged and decayed conditions. In order to ensure the reliability of these structures and the safety of the public, condition assessment, damage detection and safety evaluation is necessary. This paper presents a damage identification procedure, which is based on global change of vibration characteristics of a structure. The developed method utilises the damage index (DI) method in combination with neural network techniques to identify damage in numerical and experimental timber beam structures. The neural network ensemble approach is utilised in order to respect important diversities of different modes and to integrate individual characteristics of vibrational mode separated damage features. The method considers field testing issues associated with measurement noise, limited number of sensor arrays and environmental fluctuations. The results of damage detection using the proposed approach demonstrate its ability to determine the location and severity of all present damage cases. The outcomes show that the developed damage detection method is effective, robust and reliable.
UR - http://handle.uws.edu.au:8081/1959.7/534875
UR - http://search.informit.com.au/documentSummary;dn=848110673316859;res=IELENG
M3 - Article
SN - 1328-7982
VL - 9
SP - 181
EP - 194
JO - Australian Journal of Structural Engineering
JF - Australian Journal of Structural Engineering
IS - 3
ER -