中文摘要：

希尔伯特-黄谱可以将机械振动信号分解为有限的模式，然后以时频方式揭示机械的运行状况；然而，作为希尔伯特-黄谱的两个关键步骤，经验模式分解和希尔伯特变换都被“端点效应”所困扰；该文分析和研究了“端点效应”产生的原因，提出采用时间序列建模与预测方法对信号数据进行延拓，达到消除或改进经验模式分解和希尔伯特变换“端点效应”的目的，从而优化希尔伯特-黄谱。

英文摘要：

The Hilbert-Huang spectrum can decompose the mechanical vibration signal as finite modes ,then disclose the running status of machinery by the means of time-frequency; whereas ,both of the steps of Hilbert-Huang spectrum,the Empirical Mode Decomposition and Hilbert Transform,are disturbed by the end distorting. In this paper, the analyzsis and research on the root cause of end problem are given, and the time series modeling and predicting are introduced to extend the signal for the purpose of optimizing the EMD and HT. At last,the effect of Hilbert-Huang spectrum is optimized finally.