A modified scale-space guiding variational mode decomposition for high-speed railway bearing fault diagnosis

Rolling element bearings are broadly applied in various industrial machines, such as railway axles, gearboxes, electric motors, and turbines. Bearing fault diagnosis is important for preventing unexpected accidents and has been a hot research topic in last decades. Recently, a new signal processing method called the variational mode decomposition (VMD) has drawn substantial interest from researchers and engineers, and its applications for fault diagnosis have been widely discussed. The application of VMD can adaptively extract the IMFs and efficiently reduce the general problem of modal aliasing. The main problem of the VMD is that the number of instinct mode functions (IMFs) cannot be automatically determined, so the decomposition results are not reliable if the number of IMFs is not adequate. In this paper, a modified scale-space VMD is proposed to estimate the number of IMFs and optimize some other parameters of VMD in advance by applying scale-space representation to have a prior understand of the analysed signal and derive the parameters. A set of faulty industrial railway axle bearing signals are studied to verify the effectiveness of the proposed method. The results show that the proposed method can automatically decompose the resonance frequency bands of the faulty bearing signal and reveal the fault pattern by further conducting an envelope analysis. (C) 2018 Elsevier Ltd. All rights reserved.