基于反馈变分模式分解的单通道盲源分离算法

针对源信号数目未知时,变分模式分解(VMD)算法分离单通道混合信号的性能受人为设定的模式分量中心频率间隔Δf影响大、复杂度高的问题。提出了基于反馈机制的VMD单通道盲源分离(VMDF-SCBSS)算法;首先利用VMD将观测信号分解为两个模式分量,其次利用相似系数衡量模式分量纯度,将最纯的模式分量反馈到输入端并从输入信号中减去,最后根据所提出的由相似系数构造的循环迭代终止条件判断是否继续分解。仿真结果表明,VMDF-SCBSS算法无需人为确定Δf,能很好地分离单载波源信号并估计源信号数目,算法复杂度较低。

Single channel blind source separation algorithm based on feedback variational mode decomposition

When the number of source signals is unknown, the performance of the variational mode decomposition (VMD) algorithm for single channel mixed signal separation is greatly affected by the artificially set center frequency interval(Δ f ) of mode component, and the algorithm complexity is high. Here, the VMD algorithm based on feedback mechanism for single channel blind source separation (VMDF-SCBSS) was proposed. Firstly, the VMD algorithm was used to separate the observed signal into two mode components. Then, the similarity coefficient was used to measure the purity of mode components. The purest mode component was fed back to the input end and subtracted from the input signal. Finally, according to the proposed loop iteration termination condition constructed with similarity coefficients, whether or not continuing decomposition was judged. The simulation results showed that the VMDF-SCBSS algorithm doesn’t need Δ f preset artificially;it can be used to separate single carrier source signals, and estimate the number of source signals;its complexity is lower.