Ground Motion Effect on the Performance of a Neural Network Based Structural Health Monitoring Strategy
Structural health monitoring (SHM) and damage assessment of infrastructures is a very challenging subject. Many types of artificial neural networks have been used in SHM, among which back-propagation neural network is the most frequently used due to its high integration and powerful prediction abilities. A SHM strategy using back-propagation neural network is presented in this paper, in which two modified back-propagation neural networks were applied to detect the structural damage using direct structural responses under seismic loadings without eigenvalue analysis. In the strategy, the damage indicator was directly collected to the change of structural response under loadings, which makes on-line health monitoring possible. The proposed strategy works well under "known" earthquake records. Effect of other different earthquake records on the performance of the strategy is investigated.
W. Wang and G. Chen, "Ground Motion Effect on the Performance of a Neural Network Based Structural Health Monitoring Strategy," Proceedings of the Structures Congress (2010, Orlando, FL), pp. 3311-3320, American Society of Civil Engineers (ASCE), May 2010.
The definitive version is available at http://dx.doi.org/10.1061/41130(369)299
Structures Congress (2010: May 12-15, Orlando, FL)
Civil, Architectural and Environmental Engineering
Keywords and Phrases
Damage Detection; Earthquakes; Eigenvalues And Eigenfunctions; Structural Health Monitoring; Torsional Stress; Back Propagation Neural Networks; Damage Assessments; Earthquake Records; Eigenvalue Analysis; On-Line Health Monitoring; Structural Damages; Structural Health Monitoring (SHM); Structural Response; Neural Networks
Article - Conference proceedings
© 2010 American Society of Civil Engineers (ASCE), All rights reserved.