
Reference  Author(s)  Damage detection method 

Reference document—No data  Johnson et al. [7, 8]  Detailed Description of Phase I—Simulated 
Dyke et al. [9]  Detailed Description of Phase II—Experimental 

Phase I: Simulated data  Dyke et al. [10]  Loss of stiffness of members byoptimizing modal parameters 
Hera et al. [11]  Spikes in Level 1 details of wavelet decomposed signals 
Yang et al. [12]  Spectral analysis to identify stiffness parameters 
Hera and Hou [13]  Spikes in Level 1 details of wavelet decomposed signals 
Sun and Chang [14]  Covariance of response using wavelet packets 
Lam et al. [15]  Loss of stiffness using modal update and identification 
Yuen et al. [16]  Loss of stiffness of members using modal parameter extraction and Bayesian modal updating 
Lus et al. and Caicedo et al. [17, 18]  State space model, eigensystem realization algorithm and optimization using modal parameters 
Bernal and Gunes [19]  Extraction of a matrix proportional to structure flexibility 
Lin et al. [20]  Timefrequency features obtained using HilbertHuang transform of the intrinsic mode functions 
Chase et al. [21]  Recursive least square to identify changes in stiffness matrix 
Wu and Li [22]  Eingensensitive FE for damage detection in ambient vibration 
Yang and Huang [23]  A recursive nonlinear estimation method is used 
 Mizuno and Fujino [24]  Haar wavelet decomposition, quantization, and dissimilarity 
 Zhou et al. [25]  Residual values from subspacemodal identification 

Phase II: Simulated data  Hou and Hera [26]  Spikes in Level 1 details of wavelet decomposed signals using Daubechies and Meyer wavelets 
Barroso and Rodriguez [27]  Comparison of healthy to damage curvature in the mode shapes 
 Casciati [28]  Discrepancy between healthy and damaged states using sum of squared errors 

Phase II: Simulated and experimental data  Hera and Hou [29]  Modal parameters determined using continuous wavelet transform 
Dincal and Raich [30]  Minimization of error term between FRF of experimental & simulated data 
Nair et al. [31]  Structural stiffness change based on poles; pattern classification with autoregressive coefficients 

Phase II: experimental data only  Ching and Beck [32, 33]  ExpectationMaximization algorithm used to find most probable stiffness parameters—Config. 2–9 
Giraldo et al. [34]  Loss of stiffness of members—Config 2–6 
Lynch [35]  Pole location using system identification, Config. 1–5 
Liu et al. [36]  Timefrequency obtained using HilbertHuang transform of intrinsic modes—Config. 7 & 8 
McCuskey et al. [37] and McCuskey [38]  Neuralwavelet module—All Configurations 
Carden and Brownjohn [39]  Autoregressive moving average (ARMA) to build damage classifiers for different damage configurations 
