Stochastic Model for Simulation of Ground-Motion Sequences Using Kernel-Based Smoothed Wavelet Transform and Gaussian Mixture Distribution

Abstract

In this paper, a stochastic-parametric model is developed for simulating the temporal and spectral nonstationary characteristics of ground motion sequences. In the proposed model, after extracting the wavelet coefficients of a ground motion sequence by using the complex discrete wavelet transform and smoothing them by the Normal kernel function, they are simulated by using the Gaussian mixture distribution. This model simulates multiple peaks in the time domain, several dominant frequency peaks at each time, the relaxation time between motions, and the steps of cumulative energy curve of ground motion sequences, while the previous models did not have these abilities. © 2019, © 2019 Taylor & Francis Group, LLC.