Frequency pattern graph spectrum model and its applications in rolling bearing fault diagnosis

Abstract

Complicated working conditions and environments will intensify noise and interference in the vibration signal of the bearing, resulting in the submersion of the fault features. To enhance the weak fault information in the original signal, this paper proposes a Frequency Pattern Graph Spectrum Model (FPGS Model). The eigenvalue sequence, which concentrates the data with periodic pulse characteristics in the high order spectral band, is obtained by building the Laplacian matrix from the Fourier transform amplitude spectrum. Extracting key point based on eigenvalue sequences can refine the frequency pattern graph spectrum and reduce the computational complexity. Harmonic correlation index and sliding window are designed to mark fault features in the frequency pattern graph spectrum. This research created a collection of simulated signals to confirm the viability of the proposed method. The method was used on the bearing inner and outer ring experimental signals, and its efficacy was confirmed by contrasting it with other techniques.