论文部分内容阅读
经验模式分解可以将非线性、非平稳信号分解为有限个固有模式函数,在故障诊断中这些固有模式函数常常就是故障信号。但端点效应和分解终止条件的不当使其在分解过程中出现假频,限制了其应用。提出采用可变长极值镜像拓延法,对原信号两端包络进行拓延,有效地消除了端点效应;并提出在分解过程中采用不同的结束标准,使程序在适当的时候结束,提高了分解精度和速度。最后,将该方法应用于水轮发电机组振动信号分析中,取得了满意的效果。
Empirical mode decomposition decomposes nonlinear and non-stationary signals into a finite number of eigenmode functions, and these inherent mode functions are often fault signals in fault diagnosis. However, the improper end-point and decomposition termination conditions make it appear in the decomposition process of aliasing, limiting its application. In this paper, a variable length extreme mirror extension method is proposed to extend the envelope of both ends of the original signal, which effectively eliminates the end effect. It is also proposed that different end criteria be used in the decomposition process to make the program finish at an appropriate time, Improves resolution and speed of resolution. Finally, the method is applied to the vibration signal analysis of hydro-generator set and satisfactory results are obtained.