Application of Delayed Autocorrelation and Spectral Kurtosis in Fault Diagnosis of Rolling Bearing

Abstract

According to modulation characteristics of the fault vibration signal of rolling bearing and limitation of the traditional envelope demodulation, a new method based on delayed autocorrelation and spectral kurtosis is proposed to diagnose early fault of rolling bearing. The principle of delayed autocor-relation de-noising, delayed autocorrelation algorithms of discrete signal and the principle of spectral kurtosis which is used to select band-pass filter parameters automatically are discussed. After a rectangular window was used to intercept the unbiased estimator of autocorrelation function of the fault vibration signal, the delayed autocorrelation function was yielded by extracting the function part with a long time interval. Then by making use of the spectral kurtosis algorithm to process the delayed autocorrelation function, fast kurtosis diagram where the best band-pass filter parameters displayed was obtained, and finally the fault diagnosis can be implemented by band-pass filter and envelope demodulation. The proposed method was applied to detect the vibration signal of the outer raceway of rolling bearing, the analysis results demonstrated effectiveness of the proposed method.