Structures 29 (2021) 863–882
2352-0124/© 2020 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
Seismic response evaluation of structures using discrete wavelet transform
through linear analysis
Reza Kamgar
a
, Masoud Dadkhah
a
, Hosein Naderpour
b, *
a
Department of Civil Engineering, Shahrekord University, Shahrekord, Iran
b
Faculty of Civil Engineering, Semnan University, Semnan, Iran
A R T I C L E INFO
Keywords:
Discrete wavelet transform
OpenSees
Elastic response spectrum
Time history analysis
ABSTRACT
In this article, the discrete wavelet transform (DWT) has been used to construct the wavelet response spectrum
analysis (WRSA) of structures in the linear elastic zone. For this purpose, the elastic response spectra of different
single degree of freedom (SDOF) systems are studied for different beam-to-column stiffness ratios. Initially, using
the DWT, the records of near- and far-feld earthquakes are decomposed into three levels. Finally, the linear
response spectra of the SDOF systems are constructed subjected to the main and decomposed earthquakes. In the
end, the error values for different levels of decompositions are computed using the considered error estimation
criteria. The results show that using WRSA, the computational volume of time history analysis has reduced since
the consuming time to perform the analysis has signifcantly reduced (e.g., it reduces the computational cost by
almost 50% in the frst decomposition level). In addition, the results indicate that the highest error values occur
for the structures with a beam-to-column stiffness ratio equal to one. However, for the structures with ratios of
greater than one, the values of error reduced. For the conventional structures where the ratio is less than one and
the natural period is in the range of 0.5 s and 10 s, the error values are <10%. Results also revealed that, except
for the structures with beam-to-column stiffness ratio equal to one, in all the other cases, the values of error for
the structure subjected to the near-feld earthquakes are less than those of the far-feld ones. Furthermore, the
results obtained from the SDOF system analysis were investigated for several multi-degree-of-freedom structures
with a linear behavior.
1. Introduction
Using some mathematical transforms such as Wavelet and Fourier,
the signal information can be transferred from the time domain to do-
mains like the frequency domain or the time-frequency domain. In this
way, the information hidden in the signal can be more clearly displayed.
The wavelet transform is a newer method whose mathematical basis
dates back to the theory of Joseph Fourier in the nineteenth century. By
presenting his frequency analysis theory, Joseph Fourier showed that
each iterative function could be expressed as a fnite sum of sine and
cosine waves with different frequencies. In other words, Fourier was
able to convert a time-dependent function into a frequency-dependent
one [1]. The next step towards the current wavelet concept was taken
by Alfred Hare in 1910 [2]. In 1981, Grossman and Morlet realized that
a wavelet transform signal could be returned to its original coeffcients
without destroying its information. Grossman and Morlet contended
that they need a dual integral to convert the wavelet coeffcients back to
the original signal state since wavelets were dependent on two variables
of time and frequency. Finally, in 1984, they further realized that the
wavelet coeffcients were required to be returned with only one integral.
They also found that a small change in the wavelet coeffcients caused
only a slight change in the initial signal. The waves made of their
wavelet components were better than those generated by the Fourier
transform [3]. In 1989, Mallat showed that wavelets were implicitly
involved in the decomposition process, reducing the computational
volume of wavelet transform [4]. In addition to these, Daubechies
introduced a new group of wavelets called Daubechies wavelets using
the idea of multiple analysis. These wavelets could be implemented with
simple digital fltering and being orthogonal and orderly [5,6].
In recent years, researchers have used wavelet transforms in various
sciences related to engineering. In so doing, [7] predicted the quantity of
daily sewage sludge using different models such as wavelet-model tree
and wavelet-evolutionary polynomial regression. In another research,
[8] used the six-mother wavelet functions to increase the accuracy level
* Corresponding author at: Faculty of Civil Engineering, Semnan University, Semnan 3513119111, Iran.
E-mail address: naderpour@semnan.ac.ir (H. Naderpour).
Contents lists available at ScienceDirect
Structures
journal homepage: www.elsevier.com/locate/structures
https://doi.org/10.1016/j.istruc.2020.11.012
Received 6 August 2020; Received in revised form 6 November 2020; Accepted 9 November 2020