SVD-LMD联合降噪和TEO的滚动轴承故障诊断

针对随机噪声背景下滚动轴承局部损伤信息提取困难的问题,提出了一种奇异值分解(Singular value decomposition,SVD)和局部均值分解(Local mean decomposition,LMD)联合降噪,并结合Teager能量算子(Teager energy operator,TEO)的特征提取新方法.首先,利用SVD方法对滚动轴承故障振动信号进行处理,初步剔除背景噪声;然后,使用LMD方法分解降噪后的信号,依据相关系数指标筛分出敏感乘积函数(Product function,PF)并加以重构;最后,对重构的信号进行TEO解调分析,将解调谱中幅值突出的频率成分与故障特征频率理论值进行对比,提取故障信息.结果表明,该方法可有效提取轴承局部损伤的特征频率,最终实现故障诊断.     

基于EMD-SVD和CNN的旋转机械故障诊断

为解决旋转机械振动信号复杂且难以提取有效故障特征的问题，提出了一种经验模态分解（empirical mode decomposition，简称EMD）、奇异值分解（singular value decomposition，简称SVD）和深度卷积网络（Convolutional Neural Network，简称CNN）相结合的故障诊断方法。首先，通过EMD方法将故障信号分解成若干个固有模态分量（intrinsic mode function，简称IMF），构造时域与频域空间状态矩阵；其次，利用SVD方法对空间状态矩阵进行分解得到奇异值数组，构造时域与频域奇异值特征矩阵；最后，将提取的奇异值特征矩阵输入到CNN中进行模式识别。将该方法分别应用于滚动轴承与齿轮箱故障诊断中，在西储大学滚动轴承数据集、PHM2009直齿齿轮箱数据集上均取得了很好效果，正确率优于将原始信号直接输入到CNN中等几种对比方法，验证了该方法的优越性。     

基于形态奇异值分解和经验模态分解的滚动轴承故障特征提取方法

针对随机噪声和局部强干扰影响经验模态分解(Empirical mode decomposition,EMD)质量的问题,提出一种形态奇异值分解滤波消噪方法,并将其与EMD相结合形成一种新的故障特征提取方法。该方法首先对原始振动信号进行相空间重构和奇异值分解(Singular value decomposition,SVD),根据奇异值分布曲线确定降噪阶次进行SVD降噪,再形态滤波,最后把消噪后的信号进行EMD分解,利用本征模模态分量(Intrinsic mode function,IMF)提取故障特征信息。对仿真信号和实际轴承故障数据的应用分析表明,该方法能有效地提取轴承故障特征,诊断轴承故障,还可以减少EMD的分解层数和边界效应,提高EMD分解的时效性和精确度。     

基于IES的切削颤振孕育期信号降噪方法

切削颤振孕育期介于稳定切削与颤振爆发之间,该阶段切削力信号中颤振特征具有典型微弱信息特性。采用基于总体经验模态分解（ensemble empirical mode decomposition, 简称EEMD）与奇异值分解（singular value decomposition, 简称SVD）相结合的方法对颤振孕育期信号进行降噪时,大多存在噪声剔除不充分或微弱目标特征信息失真等问题。首先,通过引入功率谱密度（power spectral density, 简称PSD）与常相干函数（common coherency function, 简称CCF）对EEMD降噪机制进行改进,使微弱目标特征所在本征模态函数（intrinsic mode function, 简称IMF）分量得到有效提取;其次,借助池化原理（pooling principle, 简称PP）降低IMF分量复杂度,并联合SVD对其实施分块降噪,以实现对微弱目标特征中所含噪声进行有效消减;最后,耦合上述改进并重构信号,可面向微弱目标特征信号形成基于改进EEMD?SVD（improved EEMD?SVD,简称IES）的降噪方法。分别利用IES与EEMD?SVD对Rossler混沌信号进行降噪处理,并通过比较信噪比、均方误差及平滑度等降噪评价指标,对所提方法在降噪有效性及信息保真度方面的优势进行量化验证。在此基础上,再次借助所提IES方法对变轴向切深铣削实验中颤振孕育期铣削力信号进行降噪分析。结果表明,该方法能显著抑制颤振孕育期信号噪声,并能有效避免微弱颤振特征信号失真问题。     

基于SVD-SGWT和IMF能量熵增量的液压故障特征提取

针对随机噪声和虚假分量影响总体平均经验模态分解(EEMD)分解质量问题,提出基于奇异值分解(SVD)和第二代小波变换(SGWT)联合降噪预处理和本征模态分量(IMF)能量熵增量剔除虚假分量的改进EEMD方法。该方法首先对原始信号进行第二代小波变换,利用SVD对SGWT得到的高频系数进行降噪处理,克服了软、硬阈值法降噪的缺陷。然后对消噪处理的信号进行EEMD分解,通过IMF能量熵增量去除虚假分量;最后对主IMF分量进行Hilbert谱分析来提取信号的主要特征。仿真和实验结果表明,SVD和SGWT联合降噪故障信号信噪比显著提高,且失真度小,抑制了噪声对EEMD分解精度的干扰,能量熵增量能有效地去除虚假IMF,Hilbert谱中各频率成分清晰不混叠,成功提取了液压系统故障特征频率。     

基于EEMD奇异值熵的滚动轴承故障诊断方法

为充分利用振动信号进行故障辨识,提出一种基于集合经验模态分解(ensemble empirical mode decomposition,简称EEMD)奇异值熵判据的滚动轴承故障诊断方法。首先,对滚动轴承的振动信号进行EEMD分解获得若干个本征模态函数(intrinsic mode function,简称IMF),并根据一种IMF分量故障信息含量的评价指标(即峭度、均方差和欧氏距离)选出能够表征原始信号状态的分量进行信号重构;其次,利用奇异值分解技术对重构信号进行处理,结合信息熵算法求取其奇异值熵;最后,利用奇异值熵的大小判断滚动轴承的故障类别。用美国西储大学滚动轴承振动信号对所述方法进行验证的结果表明,相比传统的EMD奇异值熵故障诊断方法,本方法能够清晰的划分出滚动轴承不同工作状态的类别特征区间,而且具有更高的故障诊断精度。     

基于VMD的Volterra模型奇异值熵的转子故障诊断方法

针对转子故障信号的非平稳性以及敏感故障特征无法有效提取的问题,将变分模态分解(variational mode decomposition,VMD)的Volterra模型和奇异值熵相结合,提出一种故障诊断方法。对影响VMD分解准确性的参数选取方法进行了深入研究,给出了相关问题的解决策略。首先,对不同工况下转子实测信号进行VMD分解,利用能量熵增量选取对故障特征敏感的固有模态函数(intrinsic mode function,IMF)进行相空间重构,以建立Volterra自适应预测模型,将模型参数作为初始特征向量矩阵。然后,对初始特征向量进行奇异值分解以获取奇异值熵和奇异值特征向量矩阵,用于描述转子的故障特征。最后,采用模糊C均值(fuzzy c-means,FCM)算法对转子工作状态和故障类型进行识别。试验结果表明,所提方法可有效实现转子故障的特征提取及类型识别。通过同经集合经验模态分解(ensemble empirical mode decomposition,EEMD)相比,证明了该方法具有更有效的故障特征提取性能,是一种可行的方法。     

基于奇异值分解和局域均值分解的滚动轴承故障特征提取方法

针对随机噪声干扰滚动轴承故障特征信号提取这一问题,提出一种基于奇异值分解(Singular value decomposition,SVD)滤波降噪与局域均值分解(Local mean decomposition,LMD)相结合的故障特征提取方法。该方法首先对原始振动信号在相空间重构Hankel矩阵并利用SVD方法进行降噪处理,再对降噪后的信号进行LMD分解,将多分量的调制信号分解成一系列生产函数(Product function,PF)之和,最后结合共振解调技术对PF分量进行包络谱分析提取故障特征频率。通过数值仿真和实际轴承故障数据的分析对比,表明该方法提高了LMD的分解能力,可有效辨别出滚动轴承实测信号的典型故障,提高滚动轴承故障的诊断效果。     

基于POVMD和CAF的低转速齿轮箱故障诊断

针对低转速齿轮箱齿轮故障特征频率低、故障特征频率易被背景噪声淹没,使其难以准确提取的问题,提出了基于参数优化的变分模态分解(parameter optimization variational mode decomposition,简称POVMD)和循环自相关函数(cyclic autocorrelation function,简称CAF)结合的故障诊断方法。首先,通过POVMD对原始信号进行分解,选用余弦相似度度量选取敏感的本征模态函数(intrinsic mode function,简称IMF);其次,计算其循环自相关函数谱,获得包含调制特征的循环自相关函数谱切片;最后,使用Teager能量算子(Teager energy operator,简称TEO)算法对切片解调,提取故障特征频率。同时将本方法与相关方法进行了对比分析,特征频率提取效果更加显著,仿真信号和实验数据分析验证了该方法的有效性和可靠性。     

变量预测模型在齿轮故障诊断中的应用

将基于变量预测模型的模式识别(variable predictive model based class discriminate,简称VPMCD)方法、经验模态分解(empirical mode decomposition,简称EMD)方法和奇异值分解(singular value decomposition,简称SVD)相结合,提出了一种基于EMD,SVD和VPMCD的齿轮故障的诊断方法.首先,对齿轮振动信号进行EMD分解,得到若干个IMF(intrinsic mode function,简称IMF)分量;其次,将包含齿轮主要故障信息的前几个IMF分量组成特征向量矩阵,并对其进行SVD分解;最后,将奇异值作为特征向量建立VPMCD多故障分类器,以此来区分齿轮的工作状态和故障类型.将提出的方法应用于齿轮实验数据,分析结果表明,该方法能够实现齿轮故障类型的分类和诊断,是一种有效可行的齿轮故障诊断方法.     

Application of EEMD and improved frequency band entropy in bearing fault feature extraction

Ensemble empirical mode decomposition (EEMD) is widely used in condition monitoring of modern machine for its unique advantages. However, when the signal-to-noise ratio is low, the de-noising function of it is often not ideal. Thus, a new fault feature extraction method for rolling bearing combining EEMD and improved frequency band entropy (IFBE) is proposed, i.e., EEMD–IFBE. According to the problem of multiple intrinsic mode functions (IMFs) generated by EEMD, how to select the sensitive IMF(s) that can better reflect fault characteristics, a novel method based on FBE for sensitive IMF is proposed. In addition, since the bandwidth parameter is set empirically when the band-pass filter is designed based on the original FBE, a novel bandwidth parameter optimization method based on the principle of maximum envelope kurtosis is proposed. First, the original vibration signal is subjected to EEMD to obtain a series of IMFs; Then, the FBE values are obtained for the original signal and each IMF component, and the bandwidth of the band-pass filter (empirically) is designed as the characteristic frequency band at the minimum entropy value, and the affiliation between the characteristic frequency band of each IMF and the characteristic frequency band of the original signal is compared, and then selecting the sensitive IMF(s) that reflects the characteristics of the fault; Third, due to the influence of background noise, it is difficult to accurately obtain the fault frequency from the selected IMF(s). Therefore, the band-pass filter designed based on FBE is used, and the bandwidth parameter is optimized based on the principle of envelope kurtosis maximum, and then the selected sensitive IMF is band-pass filtered. Finally, the envelope power spectrum analysis is performed on the filtered signal to extract the fault characteristic frequency, and then the fault diagnosis of the bearing is realized. The method is successfully applied to simulated data and actual data of rolling bearing, which can accurately diagnose fault characteristics of bearing and prove the effectiveness and advantages of the method.     

基于流行-IMF奇异值熵的转子故障特征提取方法

针对转子振动信号的非平稳性以及微弱故障特征难以提取的问题，提出一种基于集合经验模式分解(ensemble empirical mode decomposition，简称EEMD)的奇异值熵和流形学习算法相结合的故障特征提取方法。首先，对原始振动信号进行EEMD分解，得到若干本征模态函数(intrinsic mode function，简称IMF)分量，根据峭度 欧式距离评价指标选取故障信息丰富的敏感分量，组成初始特征向量，求其奇异值熵；其次，利用近邻概率距离拉普拉斯特征映射算法（nearby probability distance Laplacian eigenmap,简称NPDLE）对奇异值熵组成的特征矩阵进行降维处理；最后，将得到的低维特征子集输入到K-近邻(K-nearest neighbor，简称KNN)中进行模式辨识。用一个双跨度转子实验台数据集和Iris仿真数据集对所提方法进行了验证，结果表明，IMF奇异值熵和NPDLE相结合的方法可以有效地实现转子故障特征提取，提高了故障辨识的准确性。     

全矢EEMD在轴承故障诊断中的应用

实际工况中滚动轴承故障的振动信号为非线性,非平稳的信号。为了对滚动轴承的故障做出准确识别,根据轴承故障信号的特点,在此提出一种用全矢谱和EEMD相结合来提取故障特征指标,然后利用隐马尔科夫模型对滚动轴承故障进行分类的新方法。首先对实验得到的滚动轴承同源双通道振动信号进行EEMD分解,得到数个IMF分量,选取相关性较高的分量进行全矢融合。然后提取与故障类型相对应的故障特征频率下的幅值作为滚动轴承故障分类的指标,并利用HMM方法进行训练和识别,从而区分出不同的故障类型。最后,利用实验得到的轴承故障信号进行测试,实验结果表明,该方法可以对滚动轴承故障做出较为准确的识别。     

Incipient fault detection of rolling bearing using maximum autocorrelation impulse harmonic to noise deconvolution and parameter optimized fast EEMD

Incipient Fault Detection of Rolling Bearing with heavy background noise and interference harmonics is a hot topic. In this paper, a new method based on parameter optimized fast EEMD (FEEMD) and Maximum Autocorrelation Impulse Harmonic to Noise Deconvolution (MAIHND) method is proposed for detecting the incipient fault of rolling bearing. Firstly, the FEEMD method with parameters optimization is used to reduce the noise and eliminate the interference harmonics of the fault signal. As a noise assistant improved method, the FEEMD can reduce the mode mixing and enhance the calculation efficiency significantly. Secondly, a new indicator is developed to select the sensitive IMF. Finally, a novel MAIHND method is employed to extract impulse fault feature from the sensitive IMF. Simulation and experiments results indicated that the proposed parameter optimized FEEMD–MAIHND method can effectively identify the weak impulse fault feature of rolling bearing. Moreover, the excellent performance of the proposed indicator for sensitive IMF component selection and MAIHND method is verified.     

基于EEMD和多元多尺度熵的风力发电机组滚动轴承故障特征提取

为了降低风力发电机组滚动轴承信号的噪声和进行多信道数据处理,提出了一种基于EEMD和多元多尺度熵的特征提取方法。利用EEMD算法对多信道的原始声发射信号进行分解获取无模式混淆的IMF,通过敏感度评估算法选取反应故障特征敏感的IMF进行多元多尺度熵分析,由单因素方差分析选择最优尺度对应的多元样本熵作为各种故障的特征值。通过从实验台采集得到正常、轻微损伤和断裂3种状态的样本数据,与多种特征提取方法相比较和SVM算法分类分析,证明了所选择故障特征量的准确性,同时也验证了所提出的滚动轴承故障特征提取方法的有效性和准确性。     

基于EEMD分解和奇异值差分谱理论的轴承故障诊断研究

提出一种基于总体平均经验模态分解(ensemble empirical mode decomposition,EEMD)和奇异值差分谱的轴承故障诊断方法。首先将非平稳的原始轴承振动信号通过EEMD方法分解成若干个平稳的本征模函数(intrinsic modefunction,IMF);由于背景噪声的影响,从各个IMF的频谱中难以准确地得到故障频率。对IMF分量构建Hankel矩阵,并进行奇异值分解,进一步找到奇异值差分谱,根据奇异值差分谱理论对某IMF分量进行消噪和重构,然后再求其频谱,便能准确地得到故障频率。实验结果表明,所提出的方法能有效地应用于轴承的故障诊断。     

IVMD融合奇异值差分谱的滚动轴承早期故障诊断

针对滚动轴承早期故障阶段存在特征信号微弱、故障识别相对困难的问题,提出了融合改进变分模态分解和奇异值差分谱的诊断方法。原始信号经改进变分模态分解方法处理后,被分解为若干本征模态函数分量,利用包络谱稀疏度指标筛选出最佳分量构造Hankel矩阵并进行奇异值分解,求取奇异值差分谱后,根据差分谱中的突变点重构信号,最终通过分析信号的包络谱可判断轴承的故障类型。利用改进变分模态分解融合奇异值差分谱的方法对轴承故障模拟及实测信号进行分析,均成功提取出微弱特征信息,能够实现滚动轴承早期故障的有效判别,具有一定的可靠性和应用价值。     

Incipient fault feature extraction of rolling bearings based on the MVMD and Teager energy operator

Aiming at the problems that the incipient fault of rolling bearings is difficult to recognize and the number of intrinsic mode functions (IMFs) decomposed by variational mode decomposition (VMD) must be set in advance and can not be adaptively selected, taking full advantages of the adaptive segmentation of scale spectrum and Teager energy operator (TEO) demodulation, a new method for early fault feature extraction of rolling bearings based on the modified VMD and Teager energy operator (MVMD-TEO) is proposed. Firstly, the vibration signal of rolling bearings is analyzed by adaptive scale space spectrum segmentation to obtain the spectrum segmentation support boundary, and then the number K of IMFs decomposed by VMD is adaptively determined. Secondly, the original vibration signal is adaptively decomposed into K IMFs, and the effective IMF components are extracted based on the correlation coefficient criterion. Finally, the Teager energy spectrum of the reconstructed signal of the effective IMF components is calculated by the TEO, and then the early fault features of rolling bearings are extracted to realize the fault identification and location. Comparative experiments of the proposed method and the existing fault feature extraction method based on Local Mean Decomposition and Teager energy operator (LMD-TEO) have been implemented using experimental data-sets and a measured data-set. The results of comparative experiments in three application cases show that the presented method can achieve a fairly or slightly better performance than LMD-TEO method, and the validity and feasibility of the proposed method are proved.