Quantitative Identification Of Near-fault Pulse-like Ground Motions Based On The EMD Method

Pulse-like ground motions always take place close to near-fault, mainly due to forward directivity effects or fling-step effects. These ground motions have been identified as imposing extreme demands on engineering structures, which are compared with nonpulse-like ground motions. Large number of scholars have researched about the characteristics and the effects on structures of near-fault pulse-like ground motions deeply, and quantitative identification of pulse-like ground motions has been a hot issue in recent years. At present, the majority of identification methods can only extract single pulse from velocity time history, while they are difficult to identify multi-pulse ground motions. A quantitative approach for identifying these ground motions is proposed here based on the Empirical Mode Decomposition method, the main contents and conclusions are as the follows:⑴ Velocity time history of real seismic data has performed time-frequency analysis using three approaches as follows: short time fourier transform, wavelet transform, Hilbert-Huang transform, and getting the corresponding time-frequency diagram. It is believed that Hilbert-Huang transform has obvious advantages to describe local aspects of seismic signal in comparation with the other two methods.⑵ It is necessary to classify the corresponding seismic data before using the EMD method because of the complexity of ground motion. To understand the process of identification based on the EMD clearly, related researches have carried out both single pulse and multi-pulse ground motionds. It is considered that the proposed method in this paper can not only identify single pulse but also identify multi-pulse ground motions.⑶ To verify the advantages and disadvantages of the proposed method here, this paper makes comparation with respective results of 91 known pulse-like ground motions and the selected 336 earthquake ground motion in this paper according to certain criterion using wavelet analysis and energy method and the EMD method, respectively. It is confirmed that the EMD method is the best with regard to ground motion whose the sum of the first four order energy is less than 0.3, on the contrary, the aforementioned two methods are better. As a whole, the EMD method displays advantages on the identification of multi-pulse ground motion while the other two methods are more suitable to identify single pulse.