基于伪标签半监督核局部Fisher判别分析轴承故障诊断

摘要
图/表
参考文献(15)
相关文章 (15)

全文: PDF (1425 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要针对轴承故障诊断应用中多特征融合导致的维度高、相关性强、信息冗余等问题，提出一种基于伪标签半监督核局部Fisher判别分析(Semi-supervised Kernel Local Fisher Discriminant Analysis, SS-KLFDA)轴承故障诊断方法。为了能利用大量无标签样本提高算法判别性能，该方法首先采用密度峰值聚类算法对样本进行聚类分析得到伪标签，然后通过增加规范化项到局部FDA算法的类内散度矩阵和类间散度矩阵中，以此保持无标签样本的聚类结构一致性，最后通过局部FDA算法来保持有标签样本类间散度最大化和类内散度最小化并求解最佳投影向量。为了能适应非线性数据降维，本文进一步给出了基于核的伪标签半监督局部Fisher判别算法。实验部分通过同其他流行降维算法在不同维度、不同特征集合以及不同分类算法条件下进行轴承故障诊断性能对比，结果表明本文提出的基于伪标签半监督核局部Fisher判别分析方法的分类精度明显优于其他降维算法，投影后的系数向量具有更好的区分能力，使得故障

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	陶新民
	任超
	徐朗
	何庆
	刘锐
	邹俊荣

关键词 ：故障诊断, FDA算法, 降维, 半监督

Abstract：In order to solve the problems of high-dimensionality, strong-relevance, and information redundancy due to the results of information fusion in bearing fault diagnosis domains, a novel bearing fault diagnosis approach based on semi-supervised kernel local fisher discriminant analysis (SS-KLFDA) using pseudo labels is presented in this paper. In the proposed approach, to sufficiently utilize a large amount of unlabeled samples to improve the discriminantperformance, we firstly adopt density peak clustering to acquire the pseudo cluster labels for unlabeled samples and then add regularization terms into both the local within-class and local between-class scatter matrices to preserve the local cluster structure for the unlabeled data. Finally, the added regularization terms are incorporated into the local within-class and local between-class scatter matrices concerning labeled samples to formulate the objective function and the final projection matrix are obtained by maximizing the objective function. In addition, in order to accommodate for nonlinearly separate problems, we further give a kernelized version of SS-KLFDA using pseudo labels in this study. In the experiment, the proposed approach is compared with other existing dimensionality reduction methods under different reduced dimensions, different feature sets, and different combined classifiers scenarios. The results show that the proposed approach could greatly increase the classification accuracies of all combined classifiers and possesses the best discrimination capacity among the dimension-reduced features, which can effectively improve the fault diagnosis performance.

Key words： Fault Diagnosis Fisher Discriminant Analysis Dimension Reduction Semi-Supervised

收稿日期: 2019-03-21 出版日期: 2020-08-28

引用本文:

陶新民,任超,徐朗,何庆,刘锐,邹俊荣. 基于伪标签半监督核局部Fisher判别分析轴承故障诊断[J]. 振动与冲击, 2020, 39(17): 1-9.
TAO Xinmin, REN Chao, XU Lang, HE Qing, LIU Rui, ZOU Junrong. Bearing fault diagnosis based on semi-supervised kernel local Fisher discriminant analysis using pseudo labels. JOURNAL OF VIBRATION AND SHOCK, 2020, 39(17): 1-9.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2020/V39/I17/1

[1] 宋涛,王瑜琳,赵明富,等. 基于SVDI的变工况旋转机械故障诊断方法 [J].振动与冲击, 2018, 37(19):211-216.
SONG Tao; WANG Yulin, ZHAO Mingfu, et al. Fault diagnosis for rotating machineries under variable operation conditions based on SVDI [J]. Journal of Vibration and Shock, 2018, 37(19):211-216.
[2] 沈微,陶新民. 基于同步挤压小波变换的振动信号自适应降噪方法 [J].振动与冲击, 2018, 37(14):239-247.
SHEN Wei, TAO Xinmin. Self-adaptive De-noising Algorithm for Vibration Signals Based on Synchrosqueezed Wavelet Transforms [J]. Journal of Vibration and Shock, 2018, 37(14):239-247.
[3] 李红贤,汤宝平,韩延,等. 迭代广义解调齿轮信号分离的变转速滚动轴承故障诊断 [J].振动与冲击,2018,37(23):38-44.
LI Hongxian, TANG Baoping, HAN Yan, et al. Fault diagnosis of rolling bearings under a variable rotating speed based on iterative generalized demodulation gear signal separation [J]. Journal of Vibration and Shock, 2018, 37(19):211-216.
[4] PETER N B, JOAO P H, DAVID J K. Eigenfaces vs. fisherfaces: recognition using class specific linear projection [J], IEEE Transactions Pattern Analysis Machine Intelligence. 1996, 19 (7):711–720.
[5] JIN Xiaohang, ZHAO Mingbo, CHOW T W S, et al. Motor Bearing Fault Diagnosis Using Trace Ratio Linear Discriminant Analysis [J]. IEEE Transactions on Industrial Electronics, 2013, 61(5):2441-2451.
[6] MASAAHI S, Dimensionality reduction of multimodal labeled data by local fisher discriminant analysis [J]. Journal of Machine Learning Research, 2007, 8:1027-1061.
[7] VAN M, KANG H. Bearing Defect Classification Based on Individual Wavelet Local Fisher Discriminant Analysis with Particle Swarm Optimization [J]. IEEE Transactions on Industrial Informatics, 2016, 12(1):124-135.
[8] WANG Sheng, LU Jianfeng, GU Xingjian, et al. Semi-Supervised Linear Discriminant Analysis for Dimension Reduction and Classification [J]. Pattern Recognition, 2016, 57 (C):179-189.
[9] JIANAG Li, XUAN Jianping, SHI Tielin. Feature Extraction Based on Semi-Supervised Kernel Marginal Fisher Analysis and Its Application in Bearing Fault Diagnosis [J]. Mechanical Systems and Signal Processing, 2013, 41(1-2):113-126.
[10] ZHONG Shiyong, WEN Qiaojun, GE Zhiqiang. Semi-Supervised Fisher Discriminant Analysis Model for Fault Classification in Industrial Processes [J]. Chemometrics and Intelligent Laboratory Systems, 2014, 138:203–211.
[11] GE Zhiqiang, ZHONG Shiyong, ZHANG Yingwei. Semi-supervised Kernel Learning for FDA Model and its Application for Fault Classification in Industrial Processes [J]. IEEE Transactions on Industrial Informatics, 2016, 12(4):1403-1411.
[12] RODRIGUEZ A, LAIO A. Clustering by Fast Search and Find of Density Peaks [J]. Science, 2014, 334(6191): 1492-1496.
[13] CHO H, KANG S J, KIM Y H. Image Segmentation using Linked Mean-Shift Vectors and Global/Local Attributes [J]. IEEE Transactions on Circuits & Systems for Video Technology, 2017, 27(10):2132-2140.
[14] WAN Shuting, ZHANG Xiong. Teager Energy Entropy Ratio of Wavelet Packet Transform and Its Application in Bearing Fault Diagnosis [J]. Entropy, 2018, 20(5).
[15] 李宁,杨建国,周瑞. 基于集合经验模式分解的汽油机爆震特征提取 [J]. 机械工程学报, 2015, 51(2):148-154.
LI Ning, YANG Jianguo, ZHOU Rui. Gasoline Engine Knock Feature Extraction Based on Ensemble Empirical Mode Decomposition [J]. Journal of Mechanical Engineering, 2015, 51(2):148-154.