汽轮机动静碰磨振动信号去噪方法分析研究

A Study of Turbine Rubbing Signal De-noising Technique Analysis

为确保汽轮机组故障诊断特征提取的准确性,由于现场实际振动信号掺有大量不确定噪声信号,针对现场采集的实际振动信号,进行分析去噪研究。采用小波分析、经验模态分解等方法进行分解去噪,并进行去噪对比。最后分析结果显示,基于集合经验模态分解(EEMD)的去噪方法,在对有突变故障的诊断信号自适应分解去噪弥补了经验模态分解和小波分解的缺陷,有较好的效果。

To ensure the accuracy of turbine fault diagnosis feature extraction, because of the on-site vibration signals doped a large number of uncertain noise, this paper, Analysis and de-noising the actual vibration signals with Wavelet analysis, empirical mode decomposition（EMD） and ensembl EMD （EEMD）methods to compare the effect of de-noising. Finally, the analysis showed Set-based EEMD de-noising method could adaptive decomposition and de-noising the vibration signal with mutations better, and it make up for the defect that empirical mode decomposition and wavelet decomposition.