Application of discrete wavelet transform in seismic nonlinear analysis of soil–structure interaction problems.

Simulation of soil–structure interaction (SSI) effects is a time-consuming and costly process. However, ignoring the influence of SSI on structural response may lead to inaccurate results, especially in the case of seismic nonlinear analysis. In this article, wavelet transform methodology has been utilized for investigation of the seismic response of soil–structure systems. For this purpose, different story outrigger-braced buildings resting on two different types of soil have been considered for SSI analysis. For each SSI system, several seismic records, with different values of peak ground acceleration (PGA) and peak ground velocity (PGV), have been first decomposed into approximate and detailed signals using a discrete wavelet transform. Then, seismic responses of the SSI systems subjected to the approximate signal have been evaluated. The results of the study show that, for earthquakes with low PGA/PGV ratio, the error percentage of all the parameters is smaller than 5% for the first level, and the error index is below 10% for the third level. As the PGA/PGV ratio of an earthquake increases, the concordance of approximate results with the main results decreases. However, even for the earthquakes with the PGA/PGV ratio higher than 1.2 g s/m, the first-level approximation can be used to predict seismic responses with at least 90% accuracy while significantly reducing the computational time. [ABSTRACT FROM AUTHOR]