基于Wigner分布时频遮隔的信号分解算法

综合特征值分解及Wigner分布时频遮隔提出了一种信号分解算法,并推广应用于其他交叉项抑制时频表示.对于由时频面上互不重叠分量合成的多分量信号,证明了信号分量可与各分量Wigner分布之和的逆Fourier变换的特征值分解相对应;通过阈值法可从抑制交叉项时频表示获得信号时频支撑区域,以此为模板遮隔Wigner分布可减少交叉项并保持自项聚集性,其逆Fourier变换的特征值分解就可实现多分量信号分解.仿真实例分析结果表明了该理论与算法的正确性和实用性.最后分析了算法性能并拓展了其实用范围.

Algorithm for Signal Decomposition Using Time-Frequency Masking of Wigner Distribution

Integrating the eigenvalue decomposition(ED) with time-frequency masking(TFM) of Wigner distribution(WD),an algorithm of signal decomposition was presented,and generalized to other type of TFRs with cross-terms reduction.For multi-component signals, whose components respectively exist in the time-frequency (TF) regions which does not overlapped with each other,it was testified that,signal components may be corresponded to the ED of the inverse Fourier transform (IFT) of the summation of the components’ WDs.By thresholding,the TF support regions of the WD’s auto terms maybe obtained from the TF representations (TFRs) with cross-terms reduction;and from WD by TFM,one TFR which reduces the cross-terms and keeps the auto-terms’ concentration may be obtained.Then the ED of IFT of this TFR can achieve multi-component signal decomposition.The results of simulations and instance illustrated the validity and practicability of the theory and algorithm.Finally,the performance of the algorithm was analyzed and its application was popularized.