一种抑制EEMD端点效应的改进SVR方法

针对总体平均经验模式分解(EEMD)中存在的端点效应问题,考虑到极值延拓只利用端点附近的值、基于数据点的支持向量回归机(SVR)延拓耗时长的缺点,将极值延拓和SVR数据预测延拓结合起来,提出了一种基于极值点的SVR延拓新方法。该方法通过找出信号中的所有极值点,先将端点附近的极值点作为SVR训练样本对信号极值点幅值进行延拓,并利用原信号所有相邻极值点时间尺度之差的平均值来控制延拓极值点的形状,再利用埃尔米特插值将延拓的极值点插值成所需要的数据点完成对信号的延拓。方法既参照了信号两端的变化趋势,又综合考虑了整个数据序列内部极值点的信息。通过仿真信号和对端点效应影响严重的实测液压故障信号分析表明,基于极值点的SVR延拓方法不仅提高了EEMD分解的精度和可靠性,有效地抑制了端点效应现象,解决了端点效应引起的分解失真问题,而且大幅减少了SVR延拓需要延拓的数据点,明显缩短SVR延拓的时间,提高了方法的实用性。     

基于时间序列差分特性的LMD端点效应处理方法

局部均值分解方法在计算过程中会使信号在端部产生端点效应,为减弱LMD中存在的端点效应,提出了一种基于信号时间序列差分特性的端点延拓方法。将信号曲线看作一段离散函数,利用该函数段末端的一阶差分和二阶差分的符号判断曲线的特征和发展趋势,根据具体情况判定所延拓极值点的值。使用该方法对仿真信号和实验轴承外圈故障和内圈故障信号进行了分解,并将其结果与镜像延拓改进的LMD方法分解的结果进行了对比。该方法在分解中的良好表现验证了其在信号分析中的有效性。     

HILBERT-HUANG变换端点效应处理新方法

HIBERT-HUANG变换(HHT)实现过程中,对信号进行经验模态分解(Empirical mode decomposition,EMD)和对分解得到的各个本征模函数(Intrinsic mode function,IMF)进行HILBERT变换时都会产生端点效应.对此问题,采用波形特征匹配延拓数据,提高经验模态分解精度,有效地抑制HILBERT变换中的端点效应,获得准确的HILBERT时频谱.所延拓数据兼顾原始信号中的极值点及非极值点的波形数据,使延拓数据特征与原信号一致,在HHT变换实现过程中仅需一次延拓,算法简单.仿真计算和转子系统故障试验分析结果表明,所用方法可以有效解决HHT变换的端点效应问题.     

EMD端点双极值延拓方法研究

使用经验模态分解（EMD）进行信号处理过程时，端点效应抑制是关键步骤。这里提出了一种基于包络匹配的双极值点延拓算法，实现了对EMD分解过程中端点效应的抑制处理，并通过了仿真验证。首先，将端点处连续的几个极值点的包络线作为模板子波；再者将模板子波与波形内部的极值包络进行有效匹配；然后将匹配区段前方的两个极值点作为延拓极值进行极值延拓，实现了抑制EMD分解过程中端点效应的目标。并通过仿真数据，验证了算法的有效性。     

线性外推法和相似极值法耦合抑制EMD端点效应

在应用经验模态分解(Empirical mode decomposition,EMD)进行机械故障诊断过程中,存在上下包络线拟合时产生端点效应的现象。在线性外推法和相似极值法的基础上,考虑信号内部的极值变化稳定性的情况,对EMD端点延拓进行了改进。通过计算信号内部的方差,确定信号内部稳定性的强弱。在信号内部变化平稳但稳定性较弱的情况下,采用相似极值延拓法,最大程度地保护信号内部的平稳性;在信号内部波动性较强的情况下,考虑到信号内部发生异常,采用线性外推法对信号两端进行预测延拓。模拟信号仿真和滚动轴承滚动体故障的实例分析。结果表明,该方法能有效地抑制EMD方法的端点效应。     

基于边界特征尺度匹配延拓的EMD改进方法及应用

针对经验模态分解(empirical mode decomposition,EMD)中存在的端点效应问题,提出一种边界局部特征尺度延拓的EMD改进方法。通过计算信号边缘处与信号内部任何一段子波的匹配误差,确定信号的规律性强弱。在信号内在规律性较强的情况下,进行内在的匹配波形延拓,最大限度地维护信号的内在趋势;在信号规律性较弱的情况下,特别是信号边缘发生异常变化的情况下,只考虑边缘处的局部信息,根据边缘局部极值点的特征进行延拓,对延拓的极值点序列进行包络拟合,估计出均值曲线。该方法保证了信号有效数据的正确分解,提高了信号的分解精度,实现了EMD算法的改进。仿真实验验证了该方法能较好地抑制EMD端点效应。     

Hilbert-Huang变换边界问题处理方法研究

分析EMD(empirical mode decomposition)及Hilbert变换中边界问题产生的原因,总结现有边界处理方法的特点,提出采用波形特征匹配延拓处理HHT(Hilbert-Huang transform)边界问题的新方法,该方法可以同时克服EMD分解和Hilbert变换中的端点效应问题.延拓数据来自于信号本身,兼顾原始信号极值点及非极值点的波形数据,能更好地反映信号的特征,实现延拓数据与原信号交界处的光滑过渡.在HHT计算过程中仅需一次延拓.仿真计算和故障实验分析表明,该方法可以有效解决Hilbert-huang变换中存在的边界问题.     

EMD端点效应抑制方法仿真比较与实例分析研究

基于经验模态分解的希尔伯特-黄变换分解会产生端点效应,现已提出了诸多的端点效应抑制方法。首先引入了端点效应问题,介绍了镜像延拓法、平行延拓法、极值延拓法、多项式拟合延拓法原理;提出延拓方法评价指标,采用仿真信号对四种抑制方法进行了对比分析,得出镜像延拓法是相对最优的处理方法;采用基于4种端点延拓方法改进的经验模态分解方法对凯斯西楚大学轴承故障数据依次进行处理;最终通过仿真分析与实例故障诊断均验证了镜像延拓法在解决端点效应问题方面的相对最优性。     

新型经验模式分解端点效应消除方法

对经验模式分解(Empirical mode decomposition, EMD)的端点效应进行深入分析,指出现有延拓方法的不足,对基于正弦函数延拓的方法进行改进,分析不同的延拓周期、延拓信号长度及信号端点值对EMD分解的影响,并与镜像延拓EMD分解方法进行了分析比较。在此基础上,提出一种基于指数正弦型延拓方法的EMD分解,它从原理上减小了延拓信号端点包络线的发散程度,从而提高了EMD分解精度和速度,并对其进行了仿真分析。研究结果表明,新的延拓方法优于传统的延拓方法,能较好地抑制EMD端点效应。     

基于自适应波形匹配延拓的局部均值分解端点效应处理方法

局部均值分解(LMD)能将复杂的信号自适应地分解成若干个具有物理意义的单分量信号,但是在其实现过程中会产生端点效应,使结果失真。在详细分析了LMD产生端点效应的原因后,提出了自适应波形匹配延拓法来解决端点效应。该方法充分考虑了信号的内在规律与边缘处的变化趋势,使延拓更加合理,且具自适应性。对仿真信号与实际工程信号进行了分析,结果表明该方法能有效抑制LMD分解的端点效应。     

基于极值符号序列分析的EMD端点效应处理方法

针对经验模式分解(empirical mode decomposition,简称EMD)的端点效应提出一种新的抑制方法。考虑到极值序列在EMD分解的包络线形成中占有主导地位,将信号局部极值序列进行符号化,根据符号特征进行特征匹配,在信号两端依据符号序列特征匹配结果进行符号序列拓延与对应信号还原,对拓延还原后的信号进行EMD分解以实现端点效应抑制。所提方法对于随机信号与周期信号都有着明显的抑制效果,通过对仿真信号和轴承故障信号端点效应的分析验证了方法的正确性。研究与ARMA模型、BP神经网络、镜像拓延等常见方法进行了对比,所提方法的各分量有效值指标均值为19.64%,低于其他方法,说明对低频分量有着更好的抑制效果。     

Full 3D surface reconstruction of partial scan data with noise and different levels of scale

We propose an algorithm to align partial surfaces. The algorithm does not assume specific initial positions and the same level of scaling for source and target models. The corresponding points between models are computed by detecting the scale-space extrema of the geometric invariant of the models in 2D parametric domain induced by conformal parameterization. We also propose squared distance function (SDF) to describe the invariant geometric values of a model. SDF represents the rich local shape of a surface and shows relatively robust results for a noisy model. These characteristics of SDF improve the matching result in the parametric domain for they are helpful to detect the extrema of the geometric invariant of the local shape. The proposed method computes faster the corresponding points between models and finds a greater number of corresponding points than existing methods do. Experimental results show that our proposed algorithm can successfully handle 3D scan models that have less overlap between them than that required for the existing methods.     

EMD端点效应抑制方法

针对经验模态分解(empirical mode decomposition,简称EMD)中的端点效应问题,在研究总结了现有端点效应抑制方法的基础上,提出一种新的方法——基于支持向量机(support vector machine,简称SVM)延拓和窗函数相结合的方法,弥补了SVM延拓依然找不到端点以及窗函数会改变原始信号的缺点。首先,采用SVM对原始信号两端分别进行延拓,将延拓后的数据进行加窗处理(中间加矩形窗,延拓数据加海明窗);然后,利用EMD方法对加窗后的信号进行分解,得到若干个内禀模态函数(intrinsic mode function,简称IMF);最后,将IMF分量的两端延拓部分去掉,以此来达到抑制端点效应的目的。以正交性为量化评价指标,对比分析了不同方法的性能,通过仿真和实验结果表明,该方法可以更好地抑制端点效应的发生。     

Optimization of the End Effect of Hilbert-Huang transform (HHT)

In fault diagnosis of rotating machinery, Hilbert-Huang transform(HHT) is often used to extract the fault characteristic signal and analyze decomposition results in time-frequency domain. However, end effect occurs in HHT, which leads to a series of problems such as modal aliasing and false IMF(Intrinsic Mode Function). To counter such problems in HHT, a new method is put forward to process signal by combining the generalized regression neural network(GRNN) with the boundary local characteristic-scale continuation(BLCC).Firstly, the improved EMD(Empirical Mode Decomposition) method is used to inhibit the end effect problem that appeared in conventional EMD. Secondly, the generated IMF components are used in HHT. Simulation and measurement experiment for the cases of time domain,frequency domain and related parameters of HilbertHuang spectrum show that the method described here can restrain the end effect compared with the results obtained through mirror continuation, as the absolute percentage of the maximum mean of the beginning end point offset and the terminal point offset are reduced from 30.113% and27.603% to 0.510% and 6.039% respectively, thus reducing the modal aliasing, and eliminating the false IMF components of HHT. The proposed method can effectively inhibit end effect, reduce modal aliasing and false IMF components, and show the real structure of signal components accurately.     

Application of the EMD method in the vibration analysis of ball bearings

In this paper, the empirical mode decomposition (EMD) method is introduced and the calculation of the local mean with the envelope method of the extrema is substituted by an improved method, in which the local mean is computed by using the average method of the successive extrema. Based on the improved EMD method, the vibration signals of ball bearings are analyzed in detail. The results show that the proposed method is superior to the discrete wavelet decomposition for the vibration analysis of ball bearings.     

基于特征点自动识别的B样条曲线逼近技术

提出一种实用的用三次B样条曲线逼近稠密且带噪声的二维断面数据点列的算法。剔除数据点列中的重合点并对其进行均匀弧长重采样处理后,利用相邻点拟合圆弧的方法来近似计算各数据点的离散曲率值,并根据相邻点之间的离散曲率符号变化情况及相近点之间的曲率值和曲率差分关系自动识别出断面数据中绝大多数的特征点(拐点、折痕点、曲率极值点)。构造插值于特征点的B样条曲线,并在逼近误差最大处插入新的插值点。重复这一过程,直到逼近误差小于预先给定值,从而得到最终插值点列并构造相应的B样条曲线。试验结果表明,所构造的曲线节点数目及其分布合理,能够很好地反映原始断面数据点列中的细小特征部分。该算法具有速度快、逼近精度高等特点,可广泛应用于二维断面数据的曲线重建。     

LS-SVM在基于小波变换的模态分析中端部效应的应用

系统地阐述了运用改进的Morlet小波进行模态参数识别的方法。运用小波熵对小波参数进行了优化选择从而可以进行密频模态的识别,针对小波分析时产生的端部效应问题,提出了运用最小二乘支持向量机（LS-SVM）对小波骨架进行预测延拓的方法,经预测分析后可获取较准确的模态参数。通过仿真及实验信号的验证分析,表明基于LS-SVM方法可以有效地消除端部效应,且其准确效果优于基于RBF的神经网络和时变自回归的预测方法。     

Low-pass filter without the end effect for estimating transmission characteristics—Simultaneous attaining of the end effect problem and guarantee of the transmission characteristics

A Gaussian filter (GF) is the most commonly used low-pass filter of measuring surface roughness. However, undesirable distortions, called “end effects," occur near the end points of the data ends in GF. The transmission characteristics are one of the most important indicators that show the quality of a filter. Previously, it was only possible to obtain theoretical values for the transmission characteristics of filters whose weights were given by an explicit function. In recent years, it has also become possible to obtain the transmission characteristics of a filter whose weights are given by an implicit function. However, this method has a problem in that the values near the end points of the measurement data become significantly different from each other. The consequence is that end effects may occur in the filter outputs due to side effects of the periodic extension. In the case of a spline filter (SF) applied to open profiles, the transmission characteristics of a periodic SF with an end effect can be obtained uniquely. However, the transmission characteristics of a nonperiodic SF, which has no end effect, cannot be uniquely obtained. This results in a trade-off between the two states: end effects exist in a filter whose transmission characteristics can be obtained, and the transmission characteristics of a filter without end effects cannot be uniquely obtained. To address this problem, we propose a method for the GF processing that uses shearing, point symmetric extension, and periodic extension, and produces no end effect but allows the transmission characteristics to be obtained. Previously, there was a problem with the proposed method in that the rationale was unclear regarding how to determine the reference points for point symmetric extension. We resolved this and optimized the reference points. As a result, the proposed method was shown to be successful in not only resolving the end effects in the GF, but also obtaining the transmission characteristics.     

HHT端点效应的最大Lyapunov指数边界延拓方法

针对HHT(Hilbert-Huangtransfrom)的端点效应问题,提出基于最大Lyapunov指数预测模型的HHT边界延拓方法.该方法通过相空间重构,并利用时间序列相似点的演化行为,采用最大Lyapunov指数预测模型来对时间序列的端点进行预测,有效避免了不同边界条件的三次样条插值和Hilbert变换频谱泄露对...