Rub-impact sound emission fault diagnosis method based on empirical wavelet transformation

Abstract

The invention discloses a rub-impact sound emission fault diagnosis method based on empirical wavelet transformation. The method comprises the following steps that (1) a rub-impact sound emission experimental device obtains sound emission signals; (2) the sound emission signals are divided adaptively according to spectral characteristics; (3) a wavelet window is added after division, and an empirical scale function and an empirical wavelet function are defined; (4) empirical wavelet transformation is defined; and (5) Hilbert transformation is carried out on each empirical modal component, instant frequency and amplitude are obtained, and time-frequency energy based on the empirical wavelet transformation is obtained. In the fault diagnosis method, adaptive division is carried out according to a Fourier spectrum of the sound emission signals, a wavelet filter is used to extract different inherent modal components from the sound emission signals, the analyzed modals are fewer, and phenomena as modal aliasing and terminal effects are eliminated effectively; and Hilbert transformation is carried out on the empirical modal components to obtain the instant frequency and amplitude, the time-domain energy based on empirical wavelet transformation is obtained, and fault diagnosis is carried out on the rub-impact signals.

Description

technical field [0001] The invention relates to the field of fault diagnosis of rotating machinery, in particular to a fault diagnosis method based on empirical wavelet transform and rubbing acoustic emission. Background technique [0002] Acoustic emission technology has become a research hotspot in the fault diagnosis of rotating machinery in recent years due to its wide response frequency and high sensitivity. However, acoustic emission signals are easily interfered by environmental noise, especially the large amount of interference noise generated during the operation of rotating machinery, which makes useful acoustic emission fault features blurred or even submerged, making it difficult to effectively analyze and diagnose faults. Therefore, noise reduction processing of acoustic emission signals is a prerequisite for effective diagnosis. [0003] The Fourier transform decomposes the noise while decomposing the signal, so there is a contradiction between suppressing the...

AI Technical Summary

Problems solved by technology

Once the window function in the short-time Fourier transform is selected, its time-frequency resolution is fixed, which does not meet the actual requirement that the resolution of high-frequency signals should be higher than that of low-frequency signals; although the Winger-Ville distribution has a high time-frequency resolution, but for multi-component signals, its application is greatly limited due to the existence of cross-interference items; wavelet transform has multi-resolution characteristics, but it needs to artificially select wavelet bases in applications, so it lacks Adaptive, and can not solve the problem of low frequency interference well

Empirical Mode Decomposition (EMD) can adaptively decompose non-stationary signals, but there are still some defects. It is an empirical method and lacks a complete theoretical basis. The modal obtained after decomposition The orthogonality of the components still needs to be demonstrated; the end judgment standard of the decomposition by the envelope is not scientific, which may cause the problem of modal aliasing in the decomposed signal; it takes multiple iterations to obtain the complete IMF component, Time-consuming and computationally intensive

Images