基于预处理的单通道盲源分离

Pre-processing single channel blind source separation

文中分别采用经验模态分解(EMD),变分模态分解(VMD)两种预处理方法,使单通道观测信号由欠定信号转为正定或超定从而解决其“非欠定”限制,并对此两种预处理方法进行仿真实验对比验证,研究表明,不同的预处理方法会影响信号盲分离的结果,寻找一种高效、快速的预处理方法是解决欠定盲信号分离的关键,且仿真实验结果表明,VMD预处理方法更具优越性。

This paper uses empirical mode decomposition(EMD)and variable-mode modal decomposition(VMD)two pre-processing methods to make the single-channel observation signal turn from underdetermined signal to positive or over-determined to solve its"non-underdetermined".Limitation,and the simulation experiments of the two pretreatment methods are compared and verified.The research shows that different pretreatment methods will affect the results of signal blind separation,and finding an efficient and fast preprocessing method is to solve the underdetermined blind signal separation.The key,and simulation results show that the VMD preprocessing method is more advantageous.