摘要
针对多分量机械故障振动信号的特征提取问题,介绍一种基于希尔伯特振动分解(HVD)的时频分析方法。该方法首先利用Hilbert变换得到原始振动信号的解析信号,然后通过对解析信号的瞬时频率低通滤波获得信号中幅值最大分量的瞬时频率,同时经同步检测获得相应的瞬时幅值和初相位,最后经过迭代运算自适应地检测出原信号各分量的时频信息。针对HVD方法的边界效应问题,提出一种基于相关系数准则的波形匹配边界延拓法对其进行改进。通过两组仿真信号分析验证了HVD方法对多分量非平稳信号的分解能力,同时表明改进的HVD方法能很好地抑制边界效应。给出转子系统油膜涡动故障诊断实例,验证了该方法的工程实用性。
A time-frequency approach based on Hilbert vibration decomposition （HVD）method was introduced in order to extract fault features of multi-component mechanical fault vibration signals accurately.Firstly,the analytical signals of the original vibration signals were obtained through Hilbert transformation. Secondly, the instantaneous stationary frequency of the largest amplitude component was achieved using a low-pass filtering of analytical signals＇instantaneous frequencies,the corresponding amplitude and initial phase were also estimated according to the synchronous detecting,then the time-frequency information of each component of the original signal was detected adaptively with iteration computation.Aiming at the end effects of HVD,a wave matching extending method based on correlation coefficient criteria was proposed to improve HVD.The analysis of two groups of simulated signals showed a good capacity of HVD in decomposing the non-stationary multi-component signals,and the results showed that the improved HVD suppresses its end effects.Finally,a fault diagnosis instance of an oil whirl of a rotor system was given to validate the feasibility of this method.
出处
《振动与冲击》
EI
CSCD
北大核心
2015年第3期167-171,182,共6页
Journal of Vibration and Shock
关键词
希尔伯特振动分解
多分量信号
时频分析
波形匹配延拓
机械故障诊断
Hilbert vibration decomposition
multi-component signal
time-frequency analysis
wave matching extending
mechanical fault diagnosis
