基于小波变换和极点对称模态分解对近断层多脉冲地震动的联合分析研究*

提出一种基于连续小波变换和极点对称模态分解的针对分析非线性非平稳类型的近断层多脉冲地震动信号的联合方法,并介绍了该方法的基本原理和分析优势。利用Matlab平台编制了联合分析程序并针对台站实际地震记录进行了实例分析,得到本征模态函数及瞬时频率,并与经验模态分解方法和小波分解方法分解结果对比。结果表明:(1)该方法不需要人工设定优筛分次数以及分析频率范围;(2)通过方差比率分布曲线能够自适应地确定最优筛分次数,并给出全局最优均值曲线;(3)通过直接积分法能够确定本征模态函数的瞬时频率,直接积分法和Hilbert方法的分析结果基本吻合,但直接积分方法在分析效率方面更突出;(4)实例地震动信号在得到的脉冲能量最强方向上的速度时程幅值较原记录的两个水平向速度幅值增大了50%;(5)该方法针对实例地震动在能量最强方向上抽取的脉冲分量峰值与原信号相差仅0.56%。

Conjoint Analysis of Near-fault Multi-pulse Ground Motion Based onWavelet Transformation and Extreme-point Symmetric Mode Decomposition

A new conjoint method aiming at analyzing the nonlinear non-stationary near-fault multi-pulse ground motion based on continuous wavelet transformation and extreme-point symmetric mode decomposition is proposed in this paper. The relating fundamental analysis principle and the advantage of the method are also introduced. The conjoint analysis code are implemented in Matlab and the case study are carried out on the basis of the real strong ground motions recorded at seismic stations. The empirical mode functions and the transient frequencies are obtained. And the results are analyzed and compared with those from both the empirical mode decomposition and the continuous wavelet decomposition method. The results indicate that: (1)The proposed method is characterized by the adaptive choose of the optimal sifting time and no need to setup the frequency boundary manually, which is superior to both empirical mode decomposition method and continuous wavelet transformation method; (2) The variance ratio distribution curve could be used to get the optimal sifting time adaptively, and the corresponding global optimal mean curve could also be obtained; (3) The transient frequencies of the empirical mode functions could be obtained by the direct integration method and the results from the direct integration and the Hilbert transformation agree with each other, but the former is more efficient in algorithm;(4) The peak ground velocity of the example seismic motion at the direction of maximum pulse energy increases by 50% compared with that of the original two horizontal peak ground velocities;(5) The peak value of the pulse component at the direction of maximum pulse energy extracted by the proposed method decreases 0.56% compared with the original signal at the same direction.