基于Hilbert-Huang变换理论的非平稳数据处理  

Non-stationary Data Processing Based on Hilbert-Huang Transform Theory

作　　者：程磊[1] 瞿伟廉[2] 

机构地区：[1]同济大学结构工程与防灾研究所,上海200092 [2]武汉理工大学道路桥梁与结构工程湖北省重点实验室,湖北武汉430070

出　　处：《建筑科学与工程学报》2007年第1期26-30,共5页Journal of Architecture and Civil Engineering

基　　金：国家自然科学基金项目(50378074)

摘　　要：详细介绍了Hilbert-Huang变换这一全新的处理非平稳信号数据的方法,将其用于分析土木工程中常用的El Centro地震波南北向记录。将该记录分解成9个固有模态函数分量和1个残余量,进而得到相应的瞬时频率图、瞬时能量图和振幅-频率-时间三维分布图谱,并将其边际谱与Fourier谱作了比较。结果表明,波的能量主要集中在低频分量10 Hz以下,并且Fourier谱在高频分量处高估波谱能量,在低频分量处低估波谱能量。根据此三维分布而定义的平稳度,对其频域和时域的平稳性进行了分析,得到该记录随频率及时间变化的平稳度曲线。Authors recommended the Hilbert-Huang transform which is a new method for dealing with non-stationary signal data in detail. The northern-southern record of E1 Centro seismic wave often used in civil engineering was analyzed by this method to get its instantaneous frequency map, instantaneous energy map and amplitude-frequency-time 3-D distribution by means of decomposing the record into 9 IMF components and 1 residual component. The comparison between the marginal spectrum and Fourier spectrum was also made. It is clearly shown in the comparison that the wave energy mainly concentrates on low frequency component below 10 Hz and the Fourier spectrum overestimates wave energy on high frequency component and underestimates it on low frequency component. The stationary analysis of its frequency and time domain was also made, and stationary degree curve of El Centro seismic record was gained according to the stationary degree definition defined by 3-D distribution.

关 键 词：Hilbert—Huang变换 经验模态分解 固有模态函数 非平稳 平稳度 

分 类 号：TU391[建筑科学—结构工程]