Fault diagnosis of rolling bearings using weighted horizontal visibility graph and graph Fourier transform

Abstract

Graph Fourier transform (GFT) has been proven to be an effective tool for impulse component extraction of rolling bearings, but its performance is closely related to the structure of underlying graph. Compared with the weighted path graph, the weighted horizontal visibility graph (WHVG) can reflect the dynamics characteristics of vibration signals better. When the fault bearing vibration signal is transformed into the WHVG, GFT can cluster most of the fault impulse component to the highest order range and has strong anti-interference ability. Based on WHVG and GFT, a novel fault diagnosis method for rolling bearings is proposed. In the proposed method, the graph spectrum coefficients in the highest order range are extracted to reconstruct the fault impulse component, and then the Hilbert envelope spectrum is used to diagnose the bearing fault. Simulation and experimental results demonstrate that the proposed fault diagnosis method for rolling bearings is noise tolerant and effective.