|
Damage identification to be used in a SHM system, numerous methods have been studied including methods based on modal properties (natural frequencies and mode shapes) and frequency response function (FRF). However, the methods based on modal properties may have limited capabilities for early detection of local damage due to contaminated measurements and computational error that can lead to false alarm. Besides, local damage is frequently associated with higher modes and nonlinearities and even more difficult to detect. For random, non-stationary, nonlinear, and transient signals, under pattern-level data analysis wavelet transform (WT) has been used as a signal pattern recognition method in structural damage identification. In this study, wavelet decomposition of dynamic data using modified complex Morlet wavelet with variable central frequency (MCMW+VCF) establishes a time-frequency representation for modal parameter estimation and system damage identification. Seismic response data collected from two different steel frames, a twin tower steel structure and a 3-story steel structure, under a series of both white noise and earthquake excitation were used for this discussion. First, the accuracy of MCMW+VCF for the purpose of using in the analysis of nonlinear response of structure is examined through the decomposition of vibration responses into wavelet coefficient distribution as a joint function of time and frequency. Using the wavelet coefficients, phase information and Novelty index, temporal variation of stiffness in the response of structures to strong earthquake excitation can be identified. Besides, through the total wavelet coefficient joint density function at each measurement for the pre-damage and post-damage states, detection and localization of damage can be implemented.
|
