EEMD和Cohen类联合抑制交叉项的时频特征提取方法

Time-frequency Feature Extraction Method Based on EEMD and Cohen Class to Suppress Cross Terms

Wigner-Ville分布是一种双线性时频分布,对多分量信号分析存在交叉项干扰.本文提出了一种基于EEMD和Cohen类时频融合算法,该算法采用EEMD算法将信号从频域上分离为若干个固有模态函数之和,再对伪分量之外的各分量进行Cohen类时频变换并叠加,得到信号的时频分布,消除了信号内部各模态函数之间时频分布的交叉项.通过对EEMD和Cohen类时频融合算法进行仿真,与小波分解和Cohen类联合时频算法、EMD和Cohen类联合时频算法的仿真进行比较,结果表明,该算法抑制交叉项效果最好,重构误差最小,同时抑制了噪声对时频特征的干扰.

There are cross term interference problems in multi-component signal analysis on Wigner- Ville distribution, which is a kind of bilinear time-frequency distribution. A time-frequency joint algorithm based on Ensemble Empirical Mode Decomposition （EEMD） and Cohen class was proposed. EEMD algorithm was adopted to decompose the signal into a collection of Intrinsic Mode Functions （IMF）. Cohen class time-frequency transform and superposition on each component except the pseudo one were carried out. The signal time-frequency distribution was obtained and the cross terms among IMF in internal signals were eliminated. The simulation study of time-frequency on EEMD and Cohen class was conducted. Simulation results of wavelet decomposition, Cohen class joint time-frequency algorithm and Empirical Mode Decomposition （EMD）and Cohen class joint algorithm were compared. It turned out that the EEMD and Cohen class joint algorithm has the best effect on suppressing the cross term form, diminishing the reconstruction error and restraining interference of time-frequency characteristics caused by noise.