基于LMD与Wigner-Ville分布的水力机组振动信号分析

Analysis on vibration signal of hydropower unit based on local mean decomposition and Wigner-Ville distribution

针对现有非平稳信号处理方法在分析水力机组振动信号时,无法同时兼顾较高的时频分辨率和瞬时时频特征的有效提取问题,提出了一种新的将局域均值分解(LMD)和Wigner-Ville分布相结合的非平稳信号处理方法,并将其应用于水力机组甩负荷工况的上导轴承摆度信号分析.通过与短时傅里叶变换、小波变换和Wigner-Ville分布方法提取的时频谱图对比分析,表明该方法继承了局域均值分解可以将多分量非平稳信号分解为多个频率成分单一的PF分量、以直接计算瞬时幅值和瞬时频率的优点,以及Wigner-Ville高时频分辨率的优点,同时,避免了Wigner-Ville分布因为受到固有交叉项干扰导致多个频率成分的信号难以表示清楚的缺陷,能够有效提取与转速变化趋势基本一致的瞬时时频谱图,为后续机组状态识别与故障诊断提供分析工具.

For a transient signal of a hydropower unit,the existing processing methods for non-stationary signal are unable to get a high time-frequency resolution currently,and the time-frequency characteristics cannot be extracted effectively as well. Therefore,a new method,in which the local mean decomposition( LMD) and Wigner- Ville distribution have been combined together,was proposed to analyze vibration signals of hydropower unit; further it was applied into the analysis on run-out signals in the upper guide bearing of a hydropower unit in its load rejection operation. According to comparison of the time-frequency spectra determined respectively by short-time Fourier transform,wavelet transform and Wigner- Ville distribution,the spectrum given by the present method can capture the instantaneous amplitude and frequency of a transient signal directly. This implies that the proposed method has taken the advantages of both LMD and Wigner- Ville distribution. Meanwhile,the inherent defect,i. e. crossterm interference in Wigner- Ville distribution has been removed. This new method can be used as an effective analysis tool for condition monitoring and fault diagnosis of hydropower units in future.