摘要
通过超声实验研究了沿分层介质,特别是含有低速层的分层介质传播的瑞利波频散特征.分别对有机玻璃/钢构成的两层速度递增的半空间和铝/有机玻璃/钢构成的含有低速层的三层半空间进行了超声探测实验,采用频率波数分析方法分析了多模式的瑞利波频散曲线,对于速度递增的两层半空间,得到了第一和第二个模式的频散曲线;对于含有低速层的三层介质,实验得到的频散曲线在不同频段对应不同的模式,从而在模式之间产生跳跃,分析表明模式跳跃是由各个模式在表面位移幅度的不同分布引起的.文中明确指出了低速层存在时,反演研究不仅要考虑各模式的频散特性,同时还要考虑不同模式在表面的位移分布情况,给出了低速层存在时的反演方法,避免了模式的误判.利用遗传算法对两种实验模型的介质参数进行了反演,得到了和实际参数吻合的反演结果.
The dispersive characteristics of the Rayleigh waves propagated in the layered media are investigated experimentally by ultrasonic detection test, especially in the presence of a low-velocity layer. Ultrasonic detections are performed for two layered specimens： Lucite/Steel half-space and Aluminum/Lucite/Steel halfspace. Characteristics of the multimede dispersion curves are analyzed using the frequency-wavenumber analysis method. The dispersion curves of two modes are obtained for the first specimen of a slow layer on a fast substrate. The dispersion curves for the second specimen containing a low-velocity mid-layer are discontinuous and correspond to the different modes branches due to the mode jumping. Further analysis based on the experimental data demonstrates that the mode jumping is caused by the different surface displacement distribution with frequency for each mode. This indicates that surface displacement distributions of the modes should be also considered for the inversion problems, especially in the stratified media containing a low-velocity mid-layer. Parameters of two specimens are inversed by Genetic Algorithm （GA） using the experimental dispersion curves of surface waves. The displacement distributions of each mode are considered in inversion in the presence of a low-velocity layer. Mode-misidentification is avoided using the inversion method presented in this paper and satisfied results are obtained.
出处
《地球物理学报》
SCIE
EI
CAS
CSCD
北大核心
2006年第4期1082-1091,共10页
Chinese Journal of Geophysics
基金
国家自然科学基金(10374098)
中国博士后科学基金课题资助
关键词
瑞利波反演
低速层
高阶模式
超声探测实验
遗传算法
Rayleigh wave inversion, Low-velocity layer, Higher modes, Ultrasonic detection testing, Genetic algorithm
