Applying Gaussian Process Latent Variable Model and Multi-Class Optimal Margin Distribution Machine to Fault Diagnosis
-
摘要: 轴承的振动信号特征与运行状态之间具有较强的非线性关系,导致在对轴承运行状态特征提取时,选取的高维特征间存在冗余性,因此产生故障诊断模型性能退化的问题。为此提出一种高斯过程隐变量模型(Gaussian process latent variable model,GPLVM)与多类最优边际分配机(Multi-class optimal margin distribution machine,mcODM)相结合的故障诊断方法。该方法首先对振动信号进行完备总体经验模态分解(Complementary ensemble empirical mode decomposition,CEEMD),得到信号的高维特征,并采用GPLVM对高维特征进行维数约简,然后利用约简后的特征建立mcODM故障诊断模型。轴承故障检测试验表明,该方法能够有效降低特征间的冗余性,且相较于ELM,mcODM模型能通过优化边际分布获得较高的辨识精度。Abstract: There is a strong nonlinear relationship between the vibration signal characteristics of the bearing and its running state, which leads to the redundancy among selected high-dimensional features when extracting the bearing's operating state features, thus causing the performance degradation of the fault diagnosis model. A fault diagnosis method that combines the Gaussian process latent variable model (GPLVM) with the multi-class optimal margin distribution machine (mcODM) is proposed. The method performs the complementary ensemble empirical mode decomposition (CEEMD) of the vibration signal to obtain its high-latitude characteristics and then uses the GPLVM to reduce the number of dimensions of the high-dimensional features and then uses the features that have reduced dimensions. Then it establishes the mcODM fault diagnosis model. The bearing's fault diagnosis results show that the method can effectively reduce the redundancy among the features of its vibration signals. The mcODM model can obtain higher fault diagnosis accuracy with the margin distribution optimization.
-
Key words:
- GPLVM /
- mcODM /
- CEEMD /
- dimension reduction /
- fault diagnosis /
- feature extraction
-
表 1 不同模型的辨识结果
模型 降维前的识别正确率 降维后的识别正确率 mcODM模型 85.00% 89.17% ELM模型 57.50% 84.17% 表 2 不同降维方法的mcODM模型辨识结果
方法 识别正确率 LDA 55% KPCA 86.7% -
[1] Zhang G, Yi T, Zhang T Q, et al. A multiscale noise tuning stochastic resonance for fault diagnosis in rolling element bearings[J]. Chinese Journal of Physics, 2018, 56(1):145-157 doi: 10.1016/j.cjph.2017.11.013 [2] He M, He D. Deep learning based approach for bearing fault diagnosis[J]. IEEE Transactions on Industry Applications, 2017, 53(3):3057-3065 doi: 10.1109/TIA.2017.2661250 [3] Huang N E, Shen Z, Long S R, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Proceedings of the Royal Society A:Mathematical, Physical and Engineering Sciences, 1998, 454(1971):903-995 doi: 10.1098/rspa.1998.0193 [4] Wu Z H, Huang N E. Ensemble empirical mode decomposi-tion:a noise-assisted data analysis method[J]. Advances in Adaptive Data Analysis, 2009, 1(1):1-41 http://d.old.wanfangdata.com.cn/Periodical/dianzixb201305033 [5] Yeh J R, Shieh J S, Huang N E. Complementary ensemble empirical mode decomposition:a novel noise enhanced data analysis method[J]. Advances in Adaptive Data Analysis, 2010, 2(2):135-156 doi: 10.1142/S1793536910000422 [6] Shen Z J, Chen X F, Zhang X L, et al. A novel intelligent gear fault diagnosis model based on EMD and multi-class TSVM[J]. Measurement, 2012, 45(1):30-40 doi: 10.1016/j.measurement.2011.10.008 [7] Li B, Liu P Y, Hu R X, et al. Fuzzy lattice classifier and its application to bearing fault diagnosis[J]. Applied Soft Computing, 2012, 12(6):1708-1719 doi: 10.1016/j.asoc.2012.01.020 [8] Zhao C L, Sun X B, Sun S L, et al. Fault diagnosis of sensor by chaos particle swarm optimization algorithm and support vector machine[J]. Expert Systems with Applications, 2011, 38(8):9908-9912 doi: 10.1016/j.eswa.2011.02.043 [9] D'Elia G, Cocconcelli M, Mucchi E. An algorithm for the simulation of faulted bearings in non-stationary conditions[J]. Meccanica, 2018, 53(4-5):1147-1166 doi: 10.1007/s11012-017-0767-1 [10] Liu Z H, Liu G, Pu J X, et al. Orthogonal sparse linear discriminant analysis[J]. International Journal of Systems Science, 2018, 49(9):848-858 http://d.old.wanfangdata.com.cn/NSTLQK/NSTL_QKJJ0228304873/ [11] Cheng Y J, Yuan H, Liu H M, et al. Fault diagnosis for rolling bearing based on SIFT-KPCA and SVM[J]. Engineering Computations, 2017, 34(1):53-65 doi: 10.1108/EC-01-2016-0005 [12] Wang X M, Gao X B, Yuan Y, et al. Semi-supervised Gaussian process latent variable model with pairwise constraints[J]. Neurocomputing, 2010, 73(10-12):2186-2195 doi: 10.1016/j.neucom.2010.01.021 [13] Lawrence N. Probabilistic non-linear principal component analysis with Gaussian process latent variable models[J]. Journal of Machine Learning Research, 2005, 6:1783-1816 http://cn.bing.com/academic/profile?id=963ef374aa852fa6d9180d95ece56832&encoded=0&v=paper_preview&mkt=zh-cn [14] Lawrence N D. Gaussian process models for visualisation of high dimensional data[C]//Proceedings of Neural Information Processing Systems. 2004 [15] Gao W, Zhou Z H. On the doubt about margin explanation of boosting[J]. Artificial Intelligence, 2013, 203:1-18 doi: 10.1016/j.artint.2013.07.002 [16] Zhang T, Zhou Z H. Multi-class optimal margin distribution machine[C]//Proceedings of the 34th International Conference on Machine Learning. Sydney, Australia: PMLR, 2017: 4063-4071 https://www.researchgate.net/publication/301876488_Optimal_Margin_Distribution_Machine [17] 向丹, 岑健.基于EMD熵特征融合的滚动轴承故障诊断方法[J].航空动力学报, 2015, 30(5):1149-1155 http://d.old.wanfangdata.com.cn/Periodical/hkdlxb201505016Xiang D, Cen J. Method of roller bearing fault diagnosis based on feature fusion of EMD entropy[J]. Journal of Aerospace Power, 2015, 30(5):1149-1155(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/hkdlxb201505016 [18] Lessmeier C. KAt-DataCenter: mb.uni-paderborn.de/kat/datacenter, Chair of Design and Drive Technology, Paderborn University