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

• Published in: IEEE transactions on industrial electronics (1982) Vol. 70; no. 2; pp. 1949 - 1960
• Main Authors: Miao, Yonghao, Zhang, Boyao, Li, Chenhui, Lin, Jing, Zhang, Dayi
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.02.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive filter; correlated Kurtosis (CK); Decomposition; Deconvolution; Fault diagnosis; Feature extraction; feature mode decomposition (FMD); Filter banks; Filtering theory; Finite impulse response filters; FIR filters; Impulse response; Kurtosis; Machinery; machinery fault diagnosis; Modal choice; Periodic variations; Rotating machinery; Wiener filters
• ISSN: 0278-0046; 1557-9948
• DOI: 10.1109/TIE.2022.3156156

In this article, a new decomposition theory, feature mode decomposition (FMD), is tailored for the feature extraction of machinery fault. The proposed FMD is essentially for the purpose of decomposing the different modes by the designed adaptive finite-impulse response (FIR) filters. Benefitting from the superiority of correlated Kurtosis, FMD takes the impulsiveness and periodicity of fault signal into consideration simultaneously. First, a designed FIR filter bank by Hanning window initialization is used to provide a direction for the decomposition. The period estimation and updating process are then used to lock the fault information. Finally, the redundant and mixing modes are removed in the process of mode selection. 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 simulated and experimental data collected from bearing single and compound fault. Moreover, it has been demonstrated that FMD has superiority in feature extraction of machinery fault compared with the most popular variational mode decomposition.