The Pulse-like Ground Motion Identification Based On The Singular Spectrum Analysis

It becomes important to study the characteristics and the effects of pulse-like ground motions,which have attracted lots of attentions after several severe earthquakes occurred in the 1980 s.The study of quantitative identification method for pulse-like ground motions,which aimed to obtain a large number of pulse-like ground motions,becomes a hot issue in recent years.The present identification methods mainly extracted and identified the pulse based on the hypothetical model,which is difficult to identify the multi-pulse ground motions.A quantitative pulse-like ground motion identification method was proposed based on the Singular Spectrum Analysis(SSA)to overcome the defects of the present methods in this thesis.The main contents and conclusions are as the following:(1)Several major time-frequency analysis methods,such as wavelet transform,Hilbert-Huang transform and Singular Spectrum Analysis,were compared at first,and then adopted to filter an unsteady time-history signal.The filtering results show that the SSA method is more accurate than others and SSA method also has faster calculation speed.(2)A new identification method of pulse-like ground motions based on SSA was proposed,considering the characteristics of pulses and the ideas of other methods.It contains: 1.To select areas on able database;2.To determine the approximate pulse period of the approximate single-pulse or multi-pulse ground motions by the proposed velocity response spectrum or "applicable zero method";3.To filter original signals by SSA method combing with the Hurst index;4.To extract the pulse components by the SSA method and determine the pulse period by the Fourier transform;5.To determine a reasonable identification indicator according to the ratio of singular values between the whole and extracted components.(3)The theory of wavelet analysis method,energy method and EMD method were analyzed,and a single-pulse ground motion and a multi-pulse ground motion were analyzed by wavelet analysis method,energy method and the proposed method to verify the advantages of the proposed method.And all the ground motions from the database selected in this thesis were analyzed by the wavelet analysis method,the energy method and the proposed method.It concludes that the SSA method is more effective to identify the multi-pulse ground motion,and the singular value ratio is an reasonable identification indicator with reliable precision.