A WAVELET-BASED PROCEDURE FOR MINING OF PULSE-LIKE GROUND MOTIONS FEATURES ON RESPONSE SPECTRA

The main objective of this paper is to present a wavelet-based procedure to characterize principle features of a special class of motions called pulse-like ground motions. Initially, continues wavelet transform (CWT) which has been known as a powerful technique both in earthquake engineering and seismology field is applied easily in automated detecting of strong pulse of earthquakes. In this procedure, BiorSpline (bior1.3) basis from biorthognal wavelet families is applied for extracting of the largest velocity pulse from normal-fault component of selected records. The selected wavelet decomposition process aids quantification of the effect of extracted pulses on acceleration and displacement response spectra. The results on elastic spectra show that, these pulses cause a significant amplification in the region of pulse period which should be appropriately incorporated in design procedure of structures in near field area. The results also have been presented for inelastic response spectra in different ductility level.