Selection of optimal wavelet-based damage-sensitive feature for seismic damage diagnosis

Abstract

In this research work, a more efficient wavelet-based refined damage-sensitive feature (refined DSF1) is proposed for nonlinear damage diagnosis using acceleration responses extracted from steel moment resisting frames (MRFs), which are analyzed by incremental dynamic analysis (IDA) under various ground motion record sets (140 records in total). Auto-regressive moving-average with exogenous input (ARX) method and a stabilization diagram are employed to estimate the true modal parameters from noisy modes of each MRF using power spectral density. 64 real-valued and 103 complex-valued mother wavelets considering end-effects on the wavelet coefficients are examined and the best mother wavelet-based refined DSF1 is proposed. For this purpose, Shannon entropies and coefficient of determination (R2) are used for optimal selection of central frequency (fc) and bandwidth (fb) parameters of complex Morlet (cmorfb-fc) wavelet-based refined DSF1. Comparison of the results demonstrates that the cmorfb-fc wavelet-based refined DSF1 considering both real and imaginary parts of wavelet coefficients is well correlated with the maximum story drift ratio (SDR) and has more efficiency than the Morlet wavelet-based refined DSF1, introduced in the technical literature, especially for the high-rise structures.