一种改进的Hilbert-Huang变换方法及其应用

针对希尔伯特-黄变换方法中存在的端点效应和虚假模态分量等问题，通过对上述问题产生原因的分析，提出一种基于边界极值均值延拓的端点效应抑制方法与虚假模态分量剔除算法相结合的改进方法。该方法首先根据信号两侧端点和其临近极值点的特性，对包络线在信号两端点处的位置进行约束，从而改善包络线拟合结果，在一定程度上抑制端点效应问题的产生。然后，依据各模态分量间的正交性原理，通过虚假模态分量剔除算法，选取能够反映信号特征的敏感模态分量，以保证分析的准确性。通过数值仿真，验证了所提方法的有效性。最后，应用于发动机异响故障诊断中，诊断结果表明该方法有效抑制了传统希尔伯特-黄变化方法的端点效应和虚假模态问题，可有效提取发动机异响信号的故障特征。

An Improved Hilbert-Huang Transform Method and Its Application

Aiming at the shortcoming of Hilbert-Huang transform method, where the end effect and false intrinsic mode function issue exist during its application, analyzing the reasons of these problems and an improved Hilbert-Huang transform method was proposed based on boundary extreme mean extension and false mode selecting approach. Firstly, according to the characteristics of the signal endpoint and its adjacent extreme points, estimate the value of the envelope at the two ends of the signal, and the estimated value were added to the extreme value sequence, the data extension was reasonable and the end effect can be solved. Then, the false mode selecting algorithm based on the orthogonality of each intrinsic mode function was used to select sensitive mode functions. The validity of this improved method was verified by a simulation signal analysis. Finally, the method was applied into engine abnormal sound fault diagnosis. The results showed that the improved method had a good result in dealing with the shortcoming of traditional Hilbert-Huang transform; the fault feature of engine abnormal sound signal can be obtained effectively.