Open AccessApplication of EMD and 1.5‐dimensional spectrum in fault feature extraction of rolling bearing
Author(s) -
Jiang Zhanglei,
Wu Yapeng,
Li Jun,
Liu Yaru,
Wang Jifang,
Xu Xiaoli
Publication year - 2019
Publication title -
the journal of engineering
Language(s) - English
Resource type - Journals
ISSN - 2051-3305
DOI - 10.1049/joe.2018.9121
Subject(s) - hilbert–huang transform , fault (geology) , kurtosis , bearing (navigation) , envelope (radar) , signal (programming language) , noise (video) , feature extraction , computer science , feature (linguistics) , noise reduction , algorithm , spectral density , acoustics , control theory (sociology) , pattern recognition (psychology) , mathematics , white noise , artificial intelligence , physics , statistics , telecommunications , geology , image (mathematics) , radar , linguistics , philosophy , control (management) , seismology , programming language
The original signal of rolling bearing fault contains a large number of phase coupling components and is easily submerged in the background noise, which make the fault information difficult to be extracted accurately. Aiming at the above problems, a method of fault feature extraction for rolling bearing is proposed, which combines empirical mode decomposition (EMD) with a 1.5‐dimensional spectrum. The original signal is decomposed by EMD to obtain the intrinsic modal function (IMF) of different scales. The IMF is selected by the size of a correlation coefficient and a kurtosis value to eliminate the high‐frequency components, which is reconstructed to achieve the purpose of noise reduction. The reconstructed Hilbert envelope signal is analysed by the 1.5‐dimensional spectrum to extract the nonlinear characteristic of two phase couplings, so that the fault characteristic frequency of the bearing is obtained. By analysing the signal of actual rolling bearings, the fault characteristic frequency of bearing inner and outer rings can be effectively extracted, and the validity and feasibility of the method are proved.