基于等距映射与加权KNN的旋转机械故障诊断

针对旋转机械高维复杂故障特征数据难以快速准确辨识的问题,提出一种基于等距映射非线性流形学习与加权KNN(K-nearest neighbor)分类器相结合的旋转机械故障诊断方法.在由时域统计指标和内禀模态分量能量构造的原始特征空间中,首先利用等距映射非线性流形学习算法提取旋转机械故障状态变化的本质特征,随后将提取的低维本质特征输入给加权KNN进行旋转机械的故障模式辨识.通过对齿轮箱的实验数据分析表明,该方法不仅对高维复杂的非线性故障特征具有良好的降维性能,而且故障识别率较之传统方法也明显提高,能够有效识别出高维特征空间的非线性故障特征.     

基于正交邻域保持嵌入与多核相关向量机的滚动轴承早期故障诊断

针对滚动轴承早期故障特征微弱难以快速有效辨识的问题,提出一种基于正交邻域保持嵌入(ONPE)与多核相关向量机(RVM)的滚动轴承早期故障诊断方法。首先基于多域量化特征构造表征滚动轴承早期故障的多域特征向量,基于ONPE线性流形学习对多域特征向量进行约简降维处理,获取最能反映滚动轴承早期故障运行状态变化的低维敏感特征,随后将获取的低维敏感特征输入给多核RVM进行早期故障模式的分类辨识。通过分析滚动轴承早期故障的模拟实验数据表明,该方法对高维复杂的非线性早期故障特征具有良好的约简降维性能,而且比单一核函数RVM具有更好的诊断精度。     

基于改进LLE-LSTM的化工过程故障诊断

针对化工生产过程数据多样性、高维性以及相似性的特点,传统的局部线性嵌入难以发掘数据高维非线性、不均匀特征的问题,本文提出一种改进LLE-LSTM算法。首先,运用改进LLE算法求出样本集的协方差矩阵,计算权重系数矩阵,将样本集映射到低维空间。其次,将重构的低维样本集输入LSTM模型,进一步提取样本特征。最后,对故障类型进行诊断和分类。将该方法应用到田纳西-伊斯曼（TE）过程,实验结果表明该方法具有更高的准确性和优越性。     

局部线性嵌入算法及其在信号处理中的应用

局部线性嵌入算法将高维信息通过变换降到低维数的特征空间中,从而压缩数据突出信号主要特征.该算法很好地弥补了线性降维不能发现数据集非线性结构的不足.本文详细介绍了局部线性嵌入算法的基本原理和运算步骤,并将该方法应用于混有高斯白噪声的ECG信号降噪和混有弱冲击正弦信号的特征提取中.处理结果表明,局部线性嵌入算法不仅可以处理线性信号,还能较好地处理非线性信号,具有较好地工程推广价值.     

基于监督学习的核拉普拉斯特征映射的FCM算法

监督学习的核拉普拉斯特征映射,通过非线性核映射将样本数据投影到高维核特征空间,将流形结构和样本类别信息进行有效的结合后,提取嵌入在高维数据中的低维流形特征用于模糊C-均值聚类中,这样可以有效解决高维数据集中新样本的泛化性,并且能有效提高聚类的效果。     

基于高维空间聚类的集中供热末端数据异常检测

因集中供热建筑结构、住户行为习惯等差异,末端住户供暖数据具有特征差异大、非线性强、数据量大、响应时间长等特征,在原数据空间中利用聚类分析进行异常检测造成类间数据交叉,精确性无法保证。本文提出高维高斯混合聚类算法,将数据集映射到高维空间进行聚类,利用核函数映射、内积运算与高维特征空间分解等计算方法,提高精确度,规避维数灾难。搭建工业大数据分析平台,对比K-Means、高斯混合、恒虚警率、高维高斯混合算法聚类结果与异常检测精确度,本文所提算法将准确性提高到90.72%,误报率降低到5.92%,结合该算法完成4类异常用热数据集的解释与辨识。高维高斯混合聚类可以有效分析用户用热特征、检测异常数据,辅助降低采暖能耗,实现建筑节能。     

基于ISOMAP降维的复杂轮廓异常点识别方法

高维复杂轮廓异常点识别方法研究是目前过程轮廓监控的重要课题之一。以高维复杂轮廓为研究对象,建立非参数轮廓矩阵模型,将基于测地距离的ISOMAP非线性降维技术与χ2控制图相结合,提出新的轮廓异常点识别方法,以实现高维复杂轮廓异常点的准确识别。仿真实验和实际案例的应用分析结果证实该方法在异常点识别的准确性方面具有良好的性能。     

核主元分析在透平机械状态监测中的应用

讨论了核主元分析(K erne l P rinc ipa l Com ponen t A na lys is,简称KPCA)原理,提出了基于KPCA的透平机械状态监测方法。该方法在低维特征空间利用内积核函数,实现原始空间到高维空间的非线性映射以及对高维映像数据的主元分析,从而在低维空间得到原始特征的非线性主元,并根据非线性主元构建特征子空间,实现特征提取和对透平机械状态的分类识别并监测其状态变化。对仿真数据及透平机械在正常、重负荷状态下试验数据的研究表明,KPCA分类效果比主元分析好,能有效地识别出透平机械的不同状态,并能及时监测到状态发生的变化。     

基于改进的有监督正交邻域保持嵌入的故障辨识

<正>交邻域保持嵌入(Orthogonal neighborhood preserving embedding,ONPE)是一种无监督的特征降维方法,且使用的是全局统一的邻域参数,在对高维故障特征集进行特征降维时,不能利用样本的类别标签信息和不能够根据样本空间分布的变化自适应调整邻域参数,使获得的低维特征仍出现混叠的情况。针对上述问题,提出了基于改进的有监督正交邻域保持嵌入(Improved supervised ONPE,IS-ONPE)特征降维的故障辨识方法。IS-ONPE利用样本的标签信息来调整样本点与点之间的距离以形成新的距离矩阵,通过新的距离矩阵进行邻域构建,同时利用局部集聚系数进行邻域参数的自适应调整,能够获得辨识度更高的低维特征。以低维特征作为支持向量机(Support Vector Machine,SVM)的输入来实现故障辨识。齿轮的故障辨识结果表明,所提出的方法能够提高故障辨识效果,具有一定优势。     

基于模糊C聚类的油封冷却系统鲁棒辨识研究

由于实际的复杂工业过程常常具有强非线性、不确定性、多变量以及工况变化频繁等特点,很难建立其精确的数学模型描述,使得传统控制难以达到理想的控制效果.根据目标函数选择模糊模型的结构,提出了一种非线性系统模糊建模新方法,以系统的输入和输出量试验数据为依据,确定其模糊规则,建立了系统的模糊模型.其次,将时域H_∞辨识方法应用于非线性系统的模糊建模中,使得由干扰到估计误差的最大能量增益达到最小.实例表明该方法具有一定得可行性.     

Machinery fault diagnosis using supervised manifold learning

Fault diagnosis is essentially a kind of pattern recognition. How to implement feature extraction and improve recognition performance is a crucial task. In this paper, a new supervised manifold learning algorithm (S-LapEig) for feature extraction is proposed first. Via combining preserving the consistency of local neighbor information and class labels information, S-LapEig can not only gain a perfect approximation of low-dimensional intrinsic geometric structure within the high-dimensional observation data, but also enhance local within-class relations. Based on S-LapEig, a novel fault diagnosis approach is proposed. The approach extracts the intrinsic manifold features from high-dimensional fault data by directly learning the data, and translates complex mode space into a low-dimensional feature space, in which pattern classification and fault diagnosis are carried out easily. Comparing with other feature extraction methods such as PCA, LDA and Laplacian eigenmaps, the proposed method obviously improves the classification performance of fault pattern recognition. The experiments on benchmark data and engineering instance demonstrate the feasibility and effectiveness of the new approach.     

Supervised locally linear embedding projection (SLLEP) for machinery fault diagnosis

Following the intuition that the measured signal samples usually distribute on or near the nonlinear low-dimensional manifolds embedded in the high-dimensional signal space, this paper proposes a new machinery fault diagnosis approach based on supervised locally linear embedding projection (SLLEP). The approach first performs the recently proposed manifold learning algorithm supervised locally linear embedding (SLLE) on the high-dimensional fault signal samples to learn the intrinsic embedded multiple manifold features corresponding to different fault modes, and map them into a low-dimensional embedded space to achieve fault feature extraction. For dealing with the new fault sample, the approach then applies local linear regression to find the projection that best approximates the implicit mapping from high-dimensional samples to the embedding. Finally fault classification is carried out in the embedded manifold space. The ball bearing data and rotor bed data are both used to validate the proposed approach. The results show that the proposed approach obviously improves the fault classification performance and outperform the other traditional approaches.     

Fault identification method for planetary gear based on DT-CWT threshold denoising and LE

Planetary gear is widely used in large-scale complex mechanical systems. However, because of the particularity of planetary gear transmission, serious wear and fatigue crack failures often occur in the sun gear, planet gears, and inner gear ring. In addition, every type of fault will experience different degradation processes. Improving the operation reliability of mechanical equipment through fault diagnosis of planetary gears and monitoring their degradation process is beneficial. This paper proposes a planetary gear fault identification method based on Dual-tree Complex wavelet transform (DT-CWT) threshold denoising and Laplacian eigenmaps (LE). The noise reduction processing of the original signal is achieved by the DT-CWT threshold denoising method, which takes full advantage of DT-CWT and is combined with the wavelet threshold of rigrsure principle. The original high-dimensional feature set, including the time domain features, frequency domain features, permutation entropy, and fractal box dimension of the denoised signal, is constructed from multi-angles. To solve the problems of excessively large feature dimension and the existence of redundant information, the LE algorithm is used to reduce the dimension of the original high-dimensional feature set, and the low-dimensional sensitive features are obtained. Through the above method, the effective identification of different fault states and different degradation states of the planetary gear are achieved.     

Feature extraction of kernel regress reconstruction for fault diagnosis based on self-organizing manifold learning

The feature space extracted from vibration signals with various faults is often nonlinear and of high dimension.Currently,nonlinear dimensionality reduction methods are available for extracting low-dimensional embeddings,such as manifold learning.However,these methods are all based on manual intervention,which have some shortages in stability,and suppressing the disturbance noise.To extract features automatically,a manifold learning method with self-organization mapping is introduced for the first time.Under the non-uniform sample distribution reconstructed by the phase space,the expectation maximization(EM) iteration algorithm is used to divide the local neighborhoods adaptively without manual intervention.After that,the local tangent space alignment(LTSA) algorithm is adopted to compress the high-dimensional phase space into a more truthful low-dimensional representation.Finally,the signal is reconstructed by the kernel regression.Several typical states include the Lorenz system,engine fault with piston pin defect,and bearing fault with outer-race defect are analyzed.Compared with the LTSA and continuous wavelet transform,the results show that the background noise can be fully restrained and the entire periodic repetition of impact components is well separated and identified.A new way to automatically and precisely extract the impulsive components from mechanical signals is proposed.     

Rotating machine fault diagnosis using dimension reduction with linear local tangent space alignment

A novel fault diagnosis method using dimension reduction with linear local tangent space alignment is proposed in this paper. With this method, the mixed-domain feature set is first constructed to completely characterize the property of each fault by combining Empirical Mode Decomposition (EMD) with the Autoregression (AR) model coefficients. Then, Linear Local Tangent Space Alignment (LLTSA) is used to automatically compress the high-dimensional eigenvectors of training and test samples into the low-dimensional eigenvectors which have better discrimination. By using the tangent space in the neighborhood of a data point to represent the local geometry, and aligning those local tangent spaces in the low-dimensional space (which is linearly mapped from the raw high-dimensional space), LLTSA can not only gain a perfect approximation of low-dimensional intrinsic geometric structure within the high-dimensional observation data, but can also enhance local within-class relations. Finally, the Littlewoods-Paley wavelet support vector machine (LPWSVM) is proposed to perform fault classification with the obtained low-dimensional eigenvectors. Compared with the existing methods, the proposed approach has improved the fault diagnosis precision. The experiments on deep groove ball bearings fault diagnosis demonstrated the advantage and effectiveness of the proposed fault diagnosis approach.     

基于改进ICA的工业过程故障诊断研究

如今的工业过程系统结构复杂设备精密度高,随之而来的就是系统的高故障性,所以如何准确地检测到故障的发生已经成为一大难题。该文基于提高故障诊断性能的目的,提出一种DPCA-ICA的故障诊断方法。这种方法先采用DPCA对数据进行降维和去噪处理,得到能最大反映系统信息的低维数据,然后再通过ICA方法提取独立元,进行故障诊断。仿真结果表明,改进后的ICA故障诊断方法不仅具有比传统PCA方法更低的故障误报率,并且对一些PCA难以检测的故障也有很好的诊断效果。     

基于局部平衡判别投影的旋转机械故障诊断方法

针对旋转机械故障数据的多类别、高维复杂特性导致的分类困难问题,提出一种基于局部平衡判别投影(LBDP)的故障数据集降维方法。从时域、频域和时频域多个角度提取转子振动信号的混合特征,构建原始高维故障特征集;通过LBDP选择出其中最能反映故障本质的敏感特征子集;将得到的低维特征子集输入到K近邻分类器(KNN)中进行故障模式辨识。通过一个双转子系统的振动信号集合验证了所提出方法的有效性,证明了该方法能够全面地提取出局部判别信息,使故障类别之间的差异性更清晰。     

基于KPCA的SBR过程监视

序批式反应器生化污水处理系统（SBR）具有复杂的生化反应机理，其固有的严重非线性、持续时间有限、非稳态运行等给其过程监视带来特殊困难。核主元分析（KPCA）方法通过集成算子与非线性核函数计算高维特性空间的主元成分，有效捕捉过程变量中的非线性关系。将KPCA技巧应用到序批式反应器生化污水处理系统，建立了基于KPCA的SBR污水处理过程在线监视策略。在监视暴风雨事件等典型的SBR过程异常状态时，统计指标变化灵敏，诊断及时。与线性PCA相比，显示出更高的过程监视性能。     

基于Tetrolet变换的热轧钢板表面缺陷识别方法

通过Tetrolet变换将热轧钢板表面图像分解成不同尺度和方向的子带,提取子带的Tetrolet高通系数矩阵特征,得到一个高维的特征矢量。利用核保局投影算法对高维特征矢量进行降维,将降维后的低维特征矢量输入支持向量机,从而实现热轧钢板表面缺陷的分类识别。对现场采集到的热轧钢板表面图像样本进行试验,包括横向裂纹、纵向裂纹、横向划伤、纵向划伤、结疤、麻点、网纹、压痕等8类常见热轧钢板表面缺陷,以及氧化铁皮和无缺陷等样本。试验结果表明基于Tetrolet变换方法对样本图像的识别率可达97.38%,比基于Curvelet变换、Contourlet变换等方法得到的识别率提高1%左右。