Application of Hilbert-Huang Transform in the identification of characteristics of near-fault pulse-like ground motions

• Main Authors: Zih-Rou Huang, 黃子柔
• Other Authors: Yin-Nan Huang
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/dj5ebe

碩士 === 國立臺灣大學 === 土木工程學研究所 === 107 === Pulse-type ground motions produce different characteristics and higher demands on long-period structures and have drawn great attention from academic circles in the past three decades. In recent years, many scholars have developed their own methods to identify pulse-like ground motions. Among those, Shahi and Baker (2014) used wavelet transform to separate a pulse-like velocity time series into a velocity pulse and residual time series and developed index for the identification of such type of ground motion and its pulse period. Their procedure has become the most popular one. However, the procedure has its limitation since a specific type of mother wavelet is used while the shape of velocity pulse can be very different from the mother wavelet used. Furthermore, the pulse with longer period will be prior to extract because of characteristics of the wavelet analysis. The identified velocity pulse sometimes has a very long period that is not important to most buildings. In order to avoid above disadvantages, this study introduces Hilbert Huang Transform (HHT), a time-frequency analysis tool. HHT includes two parts. The first part, Empirical Mode Decomposition (EMD), is able to decompose the seismic signal into a finite and often small number of Intrinsic Mode Functions (IMF); the second part is that converting IMFs from its time domain to a representation in the frequency and time. This study finally proposes a new method to identify pulse-like ground motions, and then comparing with the method proposed by Shahi and Baker (2014). Owing to using adaptive basis the method proposed in this study has less limitation on extracted pulse shape and therefore increases the number of pulse-like ground motions. Especially, the method is suitable to identify pulse-like ground motions having lower number of pulses. Furthermore, the new method can calculate pulse periods on time-frequency domain and identify all important frequencies from the time series.