Gearbox fault diagnosis using similarity comparison of frequency spectrum density function

GUO Yuan-jing,WEI Yan-ding,JIN Xiao-hang,YANG You-dong

DOI: https://doi.org/10.16385/j.cnki.issn.1004-4523.2018.01.019

2018-01-01

Abstract:In order to reduce the difficulty of gearbox vibration signal frequency spectrum analysis and fault identification,a gearbox fault diagnosis method based on the kernel density estimation and density function similarity comparison of frequency spectrum is proposed.Firstly,multi-set vibration signals for the gearbox under every fault condition are sampled.Then their frequency spectrum density functions are calculated by the kernel density estimation.Next,a portion of the density functions are selected,of which the arithmetic average is performed to obtain the standard density function for this kind of fault condition.Finally,according to the values of cosine similarity and correlation coefficient between the frequency spectrum density function of the test vibration signal and the standard density functions of various fault conditions,the gearbox fault corresponding to the test vibration signal will be identified.The experimental results indicate that compared to the method based on frequency spectral similarity comparison,the proposed method has higher accuracy for the gearbox fault identification.In addition,correlation coefficient shows greater differences among the different fault conditions of gearbox than cosine similarity,thus providing better applicability.