Method for extracting rolling bearing fault feature based on EEMD and K-GDE

Abstract

The invention discloses a method for extracting a rolling bearing fault feature based on EEMD and K-GDE. The method comprises the following steps: S1, measuring vibration of a rolling bear by using anacceleration sensor and a collector; S2, performing EEMD decomposition on a rolling bearing vibration signal; S3, calculating kurtosis indexes of essential modular functions; S4, calculating correlation coefficients between the essential modular functions and an original signal; S5, selecting an essential modular function with the largest kurtosis index and the largest correlation coefficient from all the essential modular functions to serve as a sensitive single component of next analysis; S6, selecting a value k from a value range, and calculating the envelop amplitude of the sensitive single component based on a k-GDE method; and S7, performing Fourier transform on the envelop amplitude to obtain a k-GDE envelop spectrum, and realizing fault diagnosis by matching a peak frequency and arolling bearing fault feature frequency. The method for extracting the rolling bearing fault feature based on the EEMD and the K-GDE has the effects of accurate and reliable algorithm, as well as simple, convenient and efficient calculation on the aspect of calculation of amplitude envelop.

Description

technical field [0001] The invention relates to the field of fault feature extraction, in particular to a rolling bearing fault feature extraction method based on EEMD and K-GDE. Background technique [0002] Rolling bearings are one of the most commonly used parts of rotating machinery, and their running quality often determines the working performance of the entire system. The vibration signal generated during the operation often has multi-component and non-stationary characteristics. At the same time, with the typical modulation effect, the spectrum shows that the modulation of the resonance frequency forms sidebands on both sides of the resonance frequency, and the sidebands in the spectrum of the actual bearing vibration signal are often is asymmetrically distributed. In addition, the multi-source noise and the instability of the vibration transmission path in the actual engineering test all bring difficulties to the fault diagnosis of rolling bearings. [0003] There...

AI Technical Summary

Problems solved by technology

However, the above methods still have some defects. First, the EMD method may cause problems such as mode aliasing and endpoint effects when decomposing multiple components. Secondly, this kind of decomposition often only selects the first decomposed essential modulus function for the next step. It does not consider the single component containing the largest fault characteristic information as the sensitive component for the next analysis from the decomposed multiple single components, so the above-mentioned various methods may cause misjudgment or even diagnosis in practical applications. Errors, etc. In addition, the specific diagnostic calculation process is quite cumbersome and inefficient

Images