基于Gabor展开的Wigner-Ville分布的交叉项消除

REMOVAL OF CROSSING-ITEMS OF WVD BASED ON GABOR EXPANSION

Wigner-Ville分布(WVD)具有最好的时频分辨率,但其固有的交叉项问题限制了其应用。根据旋转机械振动信号分析中的多分量非平稳信号的频率成分均与基频成倍数(阶比)关系,提出应用基于离散Gabor变换的迭代时变滤波和对Gabor系数的选择遮掩提取出感兴趣的各单阶分量,先分别计算各单阶分量的WVD,再将各单阶分量的WVD合并,可得到无交叉项的WVD时频谱图。仿真及实际测试试验结果表明,该方法可有效去除旋转机械多分量非平稳信号WVD分析中的交叉项干扰。

Wigner-Ville distribution(WVD) with the best time-frequency resolution is recognized,however,it is limited to applications due to its crossing-item interferences.Based on the fact that the frequency components of multi-component non-stationary signals are multiples of the fundamental frequency,so-called order,in the vibration signals of a rotating machinery,an approach is proposed to solve the problem.Exploiting the iterative time-varying filtering based on the discrete Gabor transform,the interested mono-component signals can be masked and extracted respectively.Consequently,the WVD of each component is calculated one by one.At last,the zero-crossing-item time-frequency spectrogram of WVD is obtained by merging all the WVD of the selected components.The simulations and actual test results show that this method can effectively remove the crossing-item interferences of the WVD of multi-component non-stationary signals of a rotating machinery.