Feature Mode Decomposition: New Decomposition Theory for Rotating Machinery Fault Diagnosis

Authors: Yonghao Miao; Boyao Zhang; Chenhui Li; Jing Lin; Dayi Zhang

Abstract:

Decomposition methods can separate the different components from the original signal into several regular simple modes. They are the most effective tool for signal multicomponent analysis. However, the current decomposition methods do not exhibit the excellent performance in this field since the decomposition objective and filter design are not tailored for the feature extraction of machinery fault.

Therefore, a new decomposition theory called feature mode decomposition (FMD) is proposed. Firstly, the decomposition target of FMD orientates the machinery fault feature as well as is robust to other interferences and noise. Secondly, the decomposed mode is extracted by the adaptive FIR filter. Without the restriction of the filter shape, bandwidth, and center frequency of the filter, the decomposition is more thorough.

FMD provides a new solution for decompose the machinery fault information, especially the challenging compound fault which dominates the different frequency bands. The superiority of the FMD is demonstrated to adaptively and accurately decompose the fault mode as well as robust to other interferences and noise using the real experimental data compared with the most popular VMD. FMD can be further extended to signal processing in electrical, communications, seismic, medical and other potential fields.