双谱估计结合EEMD的齿轮故障特征提取  

Gear Fault Feature Extraction Based on EEMD-Bispectrum

作　　者：吕宏强[1] 武志斐[1] 王铁[1] 谷丰收[1,2] 

机构地区：[1]太原理工大学齿轮研究所,山西太原030024 [2]哈德斯菲尔德大学计算与工程学院

出　　处：《机械设计与制造》2018年第2期217-219,223,共4页Machinery Design & Manufacture

基　　金：山西省自然科学基金资助项目(2014021024-3)

摘　　要：以实测齿轮箱振动信号为分析对象,对锥齿轮系统进行故障特征提取。通过总体平均经验模态分解(EEMD)将采集到的振动信号进行分解,对比分析原始信号功率谱密度特性和各本征模态函数(IMF)频谱特性,抽取相关频带的IMF分量进行信号重构;对重构信号利用直接法进行双谱估计,计算重构信号的双谱熵和非高斯性强度并分析其随试验时间的变化趋势。结果表明,双谱熵和非高斯性强度可以有效反映齿轮运行实时状况,可作为故障诊断和趋势预测的故障特征值。This paper researched the bevel gear fault diagnosis according to the measured gearbox vibration signals.ln this paper, the Ensemble Empirical Mode Decomposition （EEMD） was used to decompose the signals ,then reconstructed the signals using parts of the Intrinsic Mode Functions （IMFs） according to the spectral characteristics of IMFs and the Power Spectral Density （PSD） of the raw singals ;then the bispectrum estimation was calculated and analysed with the changing trend of the Bispectral Entropy and the Non-Gaussian Intensity （NGI）.The result shows that the Bispectral Entropy and the NGI can effectively reflect the real-time condition and they cart supply the efficient path for the follow-up fault diagnosis and trend prediction.

关 键 词：故障诊断 锥齿轮 双谱熵 非高斯性强度 

分 类 号：TH16[机械工程—机械制造及自动化]