An operating condition information-guided iterative variational mode decomposition method based on Mahalanobis distance criterion for surge characteristic frequency extraction of the centrifugal compressor

• Published in: Mechanical systems and signal processing Vol. 186; p. 109836
• Main Authors: Hou, Yaochun, Wu, Peng, Wu, Dazhuan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.03.2023
• Subjects: Centrifugal compressor; Early surge identification; Iterative variational mode decomposition; Mahalanobis distance; Operating condition information; Surge characteristic frequencies extraction; Iterative variational mode decomposition; Centrifugal compressor; Operating condition information; Mahalanobis distance; Early surge identification; Surge characteristic frequencies extraction
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2022.109836

•The proposed method make use of early surge formation mechanism to promote VMD.•The operating information is used to set the initial target center frequency of VMD.•The Mahalanobis distance is adopted to control the iterative VMD process.•It is needless to select the target IMF with most surge-related information.•The proposed OCIGIVMD is an effective approach for surge frequencies extraction. As a significant high-speed rotating machinery being widely used in modern industry, the centrifugal compressor is subject to a potentially damaging phenomenon called surge, which may lead to property damage, casualties and even catastrophic accidents. Therefore, it is necessary to identify the early surge state of centrifugal compressors to prevent such accidents. In this paper, an operating condition information-guided iterative variational mode decomposition (OCIGIVMD) analysis method is proposed to extract the surge characteristic frequencies (SCFs) from the vibratory signals of centrifugal compressors under early surge state. The variational mode decomposition (VMD) is an effective methodology to decompose a multi-component signal into a series of amplitude-modulated-frequency-modulated (AM-FM) sub-signals with limited bandwidth, from which the characteristic frequencies can be extracted. However, when implementing VMD, the optimal determination of the decomposition mode number is still a challenging task, especially in face of complex signals containing a lot of noise and unknown interference components. Coping with the complex vibration signals of the centrifugal compressor, the VMD in an iterative form based on operating condition information is developed to extract the SCFs, while the decomposition process is controlled by Mahalanobis distance criterion to avoid under-decomposition or over-decomposition and it is needless to select the sensitive intrinsic mode function (IMF) from the multiple IMFs generated by OCIGIVMD. Finally, comparisons with the conventional VMD, empirical mode decomposition (EMD), spectral kurtosis (SK), local mean decomposition (LMD) and two improved VMD methods are conducted using both simulation and experimental data, which demonstrates that the proposed OCIGIVMD is a superior approach in extracting SCFs of the centrifugal compressor under different operating conditions.