Generalized autocorrelation method for bearing fault feature extraction under variable speed working condition

Abstract

The invention discloses a generalized autocorrelation method for bearing fault feature extraction under a variable-speed working condition, which is combined with a period estimation method to realize balance between accumulative circulation and strong background noise and is used for bearing fault diagnosis under the variable-speed working condition. The method comprises the steps that an order tracking processing method is adopted, an original vibration signal is resampled in an order domain through instantaneous phase information, and the frequency modulation phenomenon is greatly weakened; a generalized autocorrelation method is adopted, correlation of a plurality of adjacent fragments is considered, and background noise is further weakened; on the basis that the original NRC method considers the correlation among all signal segments and cannot eliminate the influence of the accumulative periodic disturbance, only the correlation of a plurality of adjacent signal segments is considered, and the accumulation of the periodic disturbance is controlled. Compared with the traditional method, the method has the advantages that the difficulty caused by mutually restricted signal characteristics is overcome, and the effect is better.

Description

technical field [0001] The invention belongs to the field of signal processing, and in particular relates to a generalized autocorrelation method for extracting bearing fault features under variable speed operating conditions. Background technique [0002] Rolling bearings are commonly used key components in rotating machinery, and it is of great significance to detect rolling bearings as early as possible and accurately. At present, periodic detection of signals in fields such as fault diagnosis and state detection of mechanical equipment is widely used. Due to the inevitable presence of a large amount of noise in the detection environment, periodic detection under strong noise background has always been a difficult problem in signal detection. Traditional time-domain methods such as variability and anti-noise correlation (NRC) can be used for periodic detection of strong noise backgrounds, but they are only suitable for strictly periodic signals, so they cannot deal with s...

AI Technical Summary

Problems solved by technology

However, due to the existence of strong background noise, the signal transformed by ACF from the observed signal is still covered by the background noise, so it cannot suppress the strong background noise

Images