A time–frequency approach for generation of synthetic time-histories of earthquake signals

Abstract

Seismic signals are non-stationary in nature, generated by earthquake and some other sources like explosive, landslides etc. In this paper, pattern of time varying frequency components of real-time recorded data of seismic signal are investigated by computing the mean of Gabor transform. In most of the cases, it is not possible to use recorded data of strong ground motion at a given site. It is therefore essential to analyze synthetic time-history which gives the prediction of strong ground motion in the regions where actual records can’t be used. In this work, the problem has been shown in synthetic time-histories where time–frequency variation is not controlled and suggested Gabor transformation based approach for generation of site specific synthetic time-history having control on time–frequency variation. Synthetic time-histories have been generated at five different sites by computing the average of Gabor transform of eight records for different earthquakes at each site. It is also shown that shape of response spectrum calculated from the generated synthetic time-history at each site is quite comparable to shape of mean response spectrum of recorded time-histories. Synthetic time-histories of a realistic time–frequency variation will be very useful for the researchers who wish to perform the nonlinear analysis of structures for a particular site. © 2015, Akadémiai Kiadó.