Abstract: In CFST cavity detection test , Hilbert-Huang transformation was used to collect and analyze the sound , which was nonlinear and non-stationary signal. Empirical mode of the original signal was decomposed into the intrinsic mode function (IMF) components , and then Hilbert Transformation of IMF component was carried out. However , IMF components decomposed from the empirical mode decomposition may include a portion of false components. So Kolmogorov-Smimov fiuing goodness inspection test method was used to distinguish the similar probabilities of each IMF component and the original signal after decomposition. Furthermore , false components were got rid of. Finally , IMF components , whose similar probability was relatively high , were applied to get a reasonable Hilbert marginal spectrum. IMF components accurately reflect the local characteristics of the structure , and effectively realize CFST cavity detection.

Key words: Kolmogorov-Smimov inspection, Hilbert-Huang transfonmation, voice signal, probability distribution

摘要： 在钢管混凝土脱空检测试验中，采用Hilbert-Huang变换对声音这种非线性非平稳信号进行采集分析。原信号 先经验模态分解为固有模态IMF分量，然后对IMF分量进行Hilbert变换，但是经验模态分解出的IMF分量中可能 有一部分的虚假分量。利用Kolmogorov-Smirnov拟合优度检验法，来判别出分解后的每一个IMF分量与原信号概率 分布的相似概率，去除掉虚假分量，采用相似概率比较高的IMF分量，最后得出合理的Hilbert边际谱，正确反映了 结构的局部特性，有效地实现了脱空的检测。

关键词: Kolmogorov-Smimov检验法, Hilbert-Huang变换, 声音信号, 概率分布