VMD在信号分解中的K值确定方法

The K Value Determination Method of VMD in Signal Decomposition

针对变分模态分解(VMD)层数K影响分解结果准确度以及噪声信号严重影响有用信息提取的问题,提出了利用中心频率法、皮尔逊相关系数和瞬时频率均值确定K值.分别应用3种方法综合分析,选择最佳分解层数K,然后利用VMD对含噪声的信号进行自适应分解,以故障振动信号为例进行验证分析,结果表明所提方法均能够有效确定VMD中的K值.

Aiming at the problem that the number of decomposition layers K in Variational Mode Decomposition(VMD)has a great influence on the accuracy of the decomposition result and the noise contained in the signal greatly affects the extraction of useful information,this paper proposes the use of the center frequency method,the correlation coefficient method and the instantaneous frequency mean(IFM)to determine the K value.Firstly,the K value in the best VMD is selected by mean instantaneous frequency,and then the noise-containing signal is adaptively decomposed by the VMD method,and the fault vibration signal is used as an example for verification and analysis,and the results show that the method can effectively determine the K value in the VMD.