中文摘要：

经验模态分解法（EMD）的端点效应是影响该方法精度的难点问题,结合端点效应的产生原理和现有研究成果,采用镜像闭合延拓法和灰色神经网络预测法相结合的方法对信号两端的包络进行延拓;通过对仿真信号和实际信号的分析表明,该方法可以有效抑制EMD方法的端点效应.利用改进的EMD方法对提速干线铁路和客运专线铁路实测轨道不平顺信号进行研究,结果表明：京广提速干线铁路样本段轨道不平顺存在着不同程度的短波和中长波不平顺,而武广高速铁路样本段轨道不平顺主要分布于中长波区段.改进EMD方法为保障铁路安全运营提供了一种新的途径.

英文摘要：

The end effect is a fatal flaw of empirical mode decomposition （EMD）. Based on the principle of end effect and the present study situation, a method combining the mirror extension with the gray neural network is proposed to extend the data. Analysis of the simulated signal and the actual track irregularity signal shows that the proposed method restrains the end effect effectively. The track irregularity signal is studied with the improved EMD method, which collected by track recording vehicles from Beijing-- Guangzhou Railway and Wuhan--Guangzhou passenger railway. The statistic and analysis result indicates that there are relatively serious short wave and long wave track irregularity in Beijing--Guangzhou Railway, and the irregularity wavelength characteristic of Wuhan--Guangzhou passenger railway is the medium wave and long wave, Analysis of track irregularity with improved EMD is a new technical method for guaranteeing the safe running of railway.