In view of the incremental learning problem of manifold learning algorithm, an adaptive neighborhood
incremental principal component analysis (PCA) and locality preserving projection (LPP) manifold learning algorithm
is presented, and the incremental learning principle of algorithm is introduced. For incremental sample
data, the adjacency and covariance matrices are incrementally updated by the existing samples; then the dimensionality
reduction results of the incremental samples are estimated by the dimensionality reduction results of the
existing samples; finally, the dimensionality reduction results of the incremental and existing samples are updated
by subspace iteration method. The adaptive neighborhood incremental PCA-LPP manifold learning algorithm
is applied to processing of gearbox fault signals. The dimensionality reduction results by incremental learning
have very small error, compared with those by batch learning. Spatial aggregation of the incremental samples is
basically stable, and fault identification rate is increased.
DENG Shijie (邓士杰), TANG Liwei (唐力伟), ZHANG Xiaotao (张晓涛)
. Research of Adaptive Neighborhood Incremental Principal Component Analysis and Locality Preserving Projection Manifold Learning Algorithm[J]. Journal of Shanghai Jiaotong University(Science), 2018
, 23(2)
: 269
-275
.
DOI: 10.1007/s12204-018-1936-7
[1] ZHAO Z H, HAO X H. Linear locality preserving anddiscriminating projection for face recognition [J]. Journalof Electronics & Information Technology, 2013,35(2): 463-466 (in Chinese).
[2] HE X F, NIYOGI P. Locality preserving projections[C]//Neural Information Processing Systems 16. Vancouver:MIT Press, 2004: 153-160.
[3] ZHANG Z H, ZHU X Z, ZHAO J M, et al. Imageretrieval based on PCA-LPP [C]//2011 10th InternationalSymposium on Distributed Computing and Applicationsto Business, Engineering and Science. Wuxi:IEEE, 2011: 230-233.
[4] WANG J, FENG J, HAN Z Y. Locally preserving PCAmethod based on manifold learning and its applicationin fault detection [J]. Control and Decision, 2013,28(5): 683-687 (in Chinese).
[5] JIA P, YIN J S, HUANG X S, et al. IncrementalLaplacian eigenmaps by preserving adjacent informationbetween data points [J]. Pattern Recognition Letters,2009, 30(8): 1457-1463.
[6] CHIN T J, SUTER D. Out-of-sample extrapolation oflearned manifolds [J]. IEEE Transactions on PatternAnalysis and Machine Intelligence, 2008, 30(9): 1547-1556.
[7] YANG Q, CHEN G M, TONG X M, et al. Applicationof incremental local tangent space alignment algorithmto rolling bearings fault diagnosis [J]. Journal of MechanicalEigineering, 2012, 48(5): 81-86 (in Chinese).
[8] TAN C, GUAN J H, ZHOU S G. Multi-sample incrementalmanifold learning algorithm based on isogonalmapping [J]. Pattern Recognition and Artificial Intelligence,2014, 27(2): 127-133 (in Chinese).
[9] SUN J F. New method for moving object detectionbased on Gaussian kernel density estimation [J]. ComputerTechnology and Development, 2010, 20(8): 45-48(in Chinese).
[10] YAN S C, XU D, ZHANG B Y, et al. Graph embeddingand extensions: A general framework for dimensionalityreduction [J]. IEEE Transactions on PatternAnalysis and Machine Intelligence, 2007, 29(1): 40-50.
[11] LI C L, WANG Z S, JIANG H K, et al. AdaptiveHessian LLE in mechanical fault feature extraction [J].Journal of Vibration Engineering, 2013, 26(5): 758-763 (in Chinese).
