Central frequency mode decomposition and its applications to the fault diagnosis of rotating machines

To overcome current challenges in variational mode decomposition (VMD) and its variants for the fault diagnosis of rotating machines, the decomposing characteristics of two sub-models buried in VMD are thoroughly explored to seek a novel way to realize effective and adaptive signal decomposition. A central frequency mode decomposition (CFMD) is proposed based on the investigation results of their decomposing characteristics. The CFMD consists of three parts. (1) A strategy for locating effective ICFs is constructed by using the first sub-model of the VMD, where a tendency discriminant function is designed to detect ICFs with good accuracy and efficiency. (2) Through the second sub-model of the VMD, a decomposing strategy induced by the located ICFs is presented to decouple the analysis signal directly, in which the optimizing procedure and incorporating balance parameter are unneeded. (3) A dichotomy strategy for updating the bandwidth parameter is built to rapidly identify possibly decomposed results of the analysis signal provided that the expected range of bandwidth parameter is input in advance. A numerical simulation and two experimental cases validate the effectiveness of the proposed CFMD method and its superiority over some advanced methods in the fault diagnosis of rotating machines.

Automated summary

This study explores the challenges in variational mode decomposition (VMD) and its variants for the fault diagnosis of rotating machines. Based on the decomposing characteristics of two sub-models in VMD, a central frequency mode decomposition (CFMD) method is proposed. The CFMD method consists of three parts and achieves effective and adaptive signal decomposition. Numerical simulation and experimental cases validate the effectiveness and superiority of the CFMD method in the fault diagnosis of rotating machines.