Quantitative trend fault diagnosis of a rolling bearing based on Sparsogram and Lempel-Ziv

Abstract

Aiming at the misdiagnosis caused by extracting the non-fault features using Lempel-Ziv algorithm, a quantitative trend fault diagnosis method of a rolling bearing based on Sparsogram and Lempel-Ziv is proposed in this paper. The approximate linear relationships between Lempel-Ziv and the fault sizes of inner and outer races are simulated with a theoretical model of rolling bearings. Then, the fault characteristic of experimental signal of bearing inner and outer race in different severity is extracted by Sparsogram algorithm, and the frequency band is selected which reflects the fault characteristic mostly, and the Lempel-Ziv complexity index of the frequency band is calculated. This method can avoid the misdiagnosis caused by extracting non-fault features. Comparing with direct calculation of Lempel-Ziv value, the performance is much better on the quantitative diagnosis of a defective bearing. The experimental results show that the proposed method can realize the quantitative trend diagnosis on defective bearings.