期刊
IEEE SENSORS JOURNAL
卷 23, 期 3, 页码 2495-2506出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSEN.2022.3227244
关键词
Narrowband; Fault diagnosis; Rolling bearings; Spectral analysis; Sensors; Power harmonic filters; Harmonic analysis; Bearing fault diagnosis; informative frequency band (IFB); multiband fault information; weighted combined envelope spectrum
This article proposes a method for rolling bearing fault diagnosis based on a weighted combined square envelope spectrum (WCSES), which decomposes the raw signal into different narrowband signals using the empirical wavelet transform. The WCSES is obtained by weighted averaging the square envelope spectra from different decomposition levels, integrating multiband diagnostic information.
Extraction of transient repetitive impact signature caused by localized defects on rolling bearings is challenging for the condition monitoring of the mechanical system. The envelope spectrum analysis demodulating the signal over the full frequency band is a typical method for identifying the rolling bearing fault but is susceptible to broadband noise. Kurtogram and its variants performing envelope spectrum analysis after bandpass filtering around carrier frequency have extensive applications in processing signals produced by periodic mechanisms. However, these methods are designed to identify a single informative frequency band (IFB) without the ability to integrate multiband diagnostic information for bearing fault detection. Therefore, this article proposes a framework for generating a weighted combined square envelope spectrum (WCSES) for rolling bearing fault diagnosis. A 1/3-binary tree filter bank constructed by using the empirical wavelet transform is utilized to decompose the raw signal into different narrowband without losing the length of the filtered signals. The WCSES that integrates multiband diagnostic information is obtained by the weighted average of all the SESs obtained from different decomposition levels. Note that various indicators can guide the construction of WCSES under the proposed framework. In addition, a targeted indicator named generalized frequency-domain signal-to-noise ratio (GR) is employed to quantify the diagnostic information hidden in each narrowband signal. The analysis of industrial bearing fault signals validated the effectiveness and superiority of the proposed WCSES in fault feature extraction.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据