Compound fault diagnosis of rotating machinery based on OVMD and a 1.5-dimension envelope spectrum

Abstract

Owing to the character of diversity and complexity, the compound fault diagnosis of rotating machinery under non-stationary operation has turned into a challenging task. In this paper, a novel method based on the optimal variational mode decomposition (OVMD) and 1.5-dimension envelope spectrum is proposed for detecting the compound faults of rotating machinery. In this method, compound fault signals are first decomposed by using OVMD containing optimal decomposition parameters, and several intrinsic mode components are obtained. Then, an adaptive selection method based on the weight factor (WF) is presented to choose two intrinsic mode components that contain the principal fault characteristic information. Finally, the 1.5-dimension envelope spectrum of the selected intrinsic mode components is utilized to extract the compound fault characteristic information of vibration signals. The performance of the proposed method is demonstrated by using the simulation signal and the experimental vibration signals collected from a rolling bearing and a gearbox with compound faults. The analysis results suggest that the proposed method is not only capable of detecting compound faults of a bearing and a gearbox, but can separate the characteristic signatures of compound faults. The research offers a new means for the compound fault diagnosis of rotating machinery.

An Iterative Reduced KPCA Hidden Markov Model for Gas Turbine Performance Fault Diagnosis

“…Hidden Markov model is extended from Markov chains to yield inferential statistical information on a state sequence [26,27]. It comprises a finite hidden state number, and each state relates to an observation at a time point.…”

“…The log-likelihood probability (LL) is a fault index defined by the logarithm calculation to ( *| ) P X λ in Equation (27). The larger LL indicates the greater consistency of observation sequence to the HMM.…”

Mode determination in variational mode decomposition and its application in fault diagnosis of rolling element bearings

“…A hybrid model for denoising and bearing feature extraction using VMD was proposed recently by Zhang et al [22]. Further study on compound fault diagnosis and characteristic separation of bearings and gearbox signals were carried out by optimizing the VMD algorithm [23]. In the present work, an attempt has been made to study the influence of initialization and input parameters (bandwidth selection parameter, ) for the application pertaining to mechanical vibration signal processing.…”

“…It was also shown that statistical feature vectors from VMD are better than those obtained using EWT in SVM classification [20]. VMD was used in bearing fault analysis in [21][22][23] and the spreading in the frequency of the signal and intensity of vibration was demonstrated as a clear indication of a fault in bearings. VMD was also used in real-time power signal decomposition [24], brain magnetic resonance [25], ECG feature extraction and classification [26] and more recently in crude oil price analysis and forecasting [27].…”

“…In such a scenario, condition monitoring of bearing vibration signals can be said to consist of oscillatory components, impulsive signals and underlying trends caused due to system dynamics. Hence certain parameters of the VMD algorithm need to be manipulated to suit this application [18,22,23]. Moreover, the signal decomposition in VMD can be deduced as wavelet packet like filter bank structure once the center frequencies of all the modes are initialized as zero.…”