局部切空间排列算法用于轴承早期故障诊断  

Inchoate Fault Diagnosis of Rolling Bearings Based on Local Tangent Space Alignment Algorithm

作　　者：杨庆[1] 陈桂明[1] 何庆飞[1] 刘鲭洁[1] 

机构地区：[1]第二炮兵工程学院装备管理工程系,西安710025

出　　处：《振动．测试与诊断》2012年第5期831-835,867-868,共5页Journal of Vibration,Measurement & Diagnosis

摘　　要：提出了一种基于经验模态分解(EMD)和局部切空间排列算法(LTSA)相结合的滚动轴承早期故障诊断方法。首先,利用经验模态分解算法分解滚动轴承不同模式下的振动信号,得到各阶本征模态分量和残余分量,提取各分量中的幅域参数和频域参数组成原始特征参数集;然后,建立基于类别可分性测度的邻域参数k选取方法,运用局部切空间排列算法实现敏感特征提取;最后,应用该方法对滚动轴承不同状态下的振动数据进行特征提取和模式识别,对比分析改进后的局部切空间排列算法与主成分分析、核主元分析以及传统局部切空间排列算法的故障模式识别能力。分析结果表明,该方法提取的滚动轴承故障特征敏感性较好,提高了故障模式识别能力,实现了滚动轴承的早期故障诊断。The method of feature extraction approach based on the empirical mode decomposition (EMD) and the local tangent space alignment algorithm(LTSA) is proposed. The vibration signal of rolling bearing in different is decomposed into a series of intrinsic mode functions (IMF) and residue component decomposed by EMD method. The original feature space with the amplitude domain parameters and the frequency domain parameters construct from a series of IMF and a residue component, respectively. Establish appraisement method of local neighborhood parameter of k based on class separability measurement, and use the local target space alignment algorithm to extract low dimensional sensitive feature. The vibration signal of rolling bearing in different mode is analyzed by the proposed method, and the ability of pattern recognition using the proposed method is compared with principal component analysis(PCA), kernel principal component analysis(KPCA) and traditional LTSA. Experimental results show that, the method can extract the sensitive fault feature from the vibration signal of the rolling bearing, and improve the ability of pattern recognition and fault diagnosis for the inchoate fault of the rolling bearing.

关 键 词：特征提取 局部切空间排列算法 经验模态分解 模式识别 滚动轴承 

分 类 号：TH17[机械工程—机械制造及自动化] TH113.1