Open AccessResearch on gear fault degree recognition method based on multi sensor fusion
Author(s) -
Xihui Chen
Publication year - 2020
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/892/1/012088
Subject(s) - pattern recognition (psychology) , kurtosis , aliasing , fault (geology) , computer science , artificial intelligence , sensor fusion , fusion , entropy (arrow of time) , artificial neural network , wavelet , time domain , signal (programming language) , mathematics , computer vision , statistics , quantum mechanics , seismology , undersampling , programming language , geology , linguistics , philosophy , physics
The fatal fault of gear will inevitably experience the evolution of different fault degrees, and the accurate recognition of the fault degree of gear has more practical significance for predictive maintenance and efficient operation. A gear fault degree recognition method based on multi sensor fusion is proposed. Ensemble local mean decomposition (ELMD) is used to decompose the vibration signal of gear, which can eliminate the aliasing effect of LMD. Then, the fault feature is extracted by using envelope spectrum information entropy and time-domain kurtosis. Based on the initial recognition of the sub evidence formed by a signal sensor based on wavelet neural network (WNN), the basic belief function assignment and fusion methods of Dempster-Shafer (D-S) evidence theory are determined, and the accurate recognition based on multi sensor fusion for different fault degree of gear is realized, which is verified by experiments.