Machinery fault diagnosis using supervised manifold learning

Detection of weak transient signals based on wavelet packet transform and manifold learning for rolling element bearing fault diagnosis

“…The technique can be realized through several algorithms including locally linear embedding (LLE) [17], isometric feature mapping (IsoMap) [18], local tangent space alignment (LTSA) [19], and Laplacian eigenmaps (LE) [20], etc. Many studies have been conducted by applying manifold learning to the machinery fault diagnosis [21][22][23][24][25][26][27][28][29][30][31][32]. In recent years, the application of manifold learning in mechanical fault diagnosis can be divided into two categories.…”

“…In recent years, the application of manifold learning in mechanical fault diagnosis can be divided into two categories. On one hand, manifold learning is used to extract non-linear features for fault classification [21][22][23][24][25][26][27][28], on the other hand, this technique is applied to extract the transient signals from the noise contaminated signal [29][30][31][32]. The above mentioned studies have demonstrated that manifold learning is effective to extract the intrinsic manifold features related to non-linear dynamics of the mechanical system.…”

Life grade recognition of rotating machinery based on Supervised Orthogonal Linear Local Tangent Space Alignment and Optimal Supervised Fuzzy C-Means Clustering

“…Recently, a novel theory of nonlinear dimension reduction called manifold learning is becoming the research focus. It can discover the intrinsic feature of nonlinear high-dimensional data by projecting 4 them into a low-dimensional feature space and preserving the local neighborhood structure [15][16][17][18]. Typical manifold learning methods mainly include Linear Discriminate Analysis (LDA) [19], Neighborhood Preserving Embedding (NPE) [20], Locality Preserving Projection (LPP) [21], Linear Local Tangent Space Alignment (LLTSA) [22], etc.…”

Modified kernel principal component analysis based on local structure analysis and its application to nonlinear process fault diagnosis

“…Specifically, Zhang et al [29] presented an LLE-based sensor fault detection method, while Li and Zhang [30] proposed to use the supervised LLE projection for machinery fault diagnosis. Jiang and co-authors [31,32] applied the modified LE method for fault pattern classification, while Hu and Yuan [33] proposed the multiway LPP (MLPP) method for batch process monitoring and demonstrated that the MLPP outperforms the conventional multiway PCA. Shao et al [34] introduced a nonlinear fault diagnosis based on the generalized LPP method which imposes orthogonality constraints on the projection vectors, while Yu [35,36] used the LPP for bearing performance degradation assessment, combing with an exponential weighted moving average statistic and Gaussian mixture models, respectively.…”