转子系统故障数据集D-LLE降维方法研究

石坤举; 赵荣珍

转子系统故障数据集D-LLE降维方法研究

石坤举^1,2,
赵荣珍^1,2

1.
兰州理工大学数字制造技术与应用省部共建教育部重点实验室, 兰州 730050;
2.
兰州理工大学机电工程学院, 兰州 730050

基金项目:

国家自然科学基金项目(50875118,51165019)

兰州理工大学博士基金项目(SB02200702)资助

详细信息

作者简介:
石坤举(1984－)，硕士，研究方向为旋转机械故障特征提取，shikunjv@sina.cn;赵荣珍(联系人)，教授，博士生导师，zhaorongzhen@lut.cn

石坤举(1984－)，硕士，研究方向为旋转机械故障特征提取，shikunjv@sina.cn;赵荣珍(联系人)，教授，博士生导师，zhaorongzhen@lut.cn

计量
- 文章访问数: 177
- HTML全文浏览量: 18
- PDF下载量: 4
- 被引次数: 0
出版历程
- 收稿日期: 2012-10-20
- 刊出日期: 2015-06-10

D-LLE Data Set Dimensionality Reduction Method in Rotor System Failure

Shi Kunju^1,2,
Zhao Rongzhen^1,2

1.
Key Laboratory of Digital Manufacturing Technology and Application, The Ministry of Education, Lanzhou University of Technology, Lanzhou 730050;
2.
College of Mechano-electronic Engineering, Lanzhou University of Technology, Lanzhou 730050

摘要

摘要: 针对传统降维方法中存在丢失判别信息及由高维空间原始特征张成的超曲面曲率较大时难以获取低维敏感信息的问题,提出一种基于Dijkstra算法的改进LLE(local linear embedding)转子故障数据集降维方法,即D-LLE法。在由时域、频域组成的原始特征空间中,利用Dijkstra算法具有可细致刻画出由时域、频域组成的原始特征空间的能力,结合LLE算法具备能够保持降维前后的转子故障数据集其流形保持不变的性质,据此可提取出反映转子运行状态的低维敏感特征属性。转子实验台模拟出的4种运行状态进行试验表明:优化后的特征数据集具有较好的聚类与类间可分性。
- 低维敏感信息 /
- LLE算法 /
- Dijkstra算法 /
- 原始特征空间
Abstract: Traditional feature extraction methods usually exist two shortcomings: discriminate information is lost and the low dimensional sensitive information is hard to obtain when the original high dimension space possesses a larger hyper surface curvature. To solve these problems,D-LLE data set dimensionally reduction method is proposed which based on the Dijkstra and Local Linear Embedding (LLE)for rotor fault feature extracting. This new method is used in original feature space that consists of time domain information and frequency domain information to extract low dimensional sensitive information. Dijkstra algorithm which has the ability of specifically depicted the original feature space and the LLE algorithm has the nature of keeping manifold invariant during the dimensionality reduc-tion. So it can extract sensitive characteristics which reflect the running state of rotor machine effectively. Experi-mental results on the rotor experimental table show that the final features extracted by our proposed method are more suitable for clustering and classification applications.
- algorithms /
- classification of information /
- cluster analysis /
- design

HTML全文

参考文献(15)

[1]	张超，陈建军，郭讯．基于 EEMD 能量熵和支持向量机的齿轮故障诊断方法[J]. 中南大学学报，2012，43(3): 932-938 Zhang C，Chen J J，Guo X. Gear fault diagnosis method based on ensemble empirical mode decomposition energy entropy and support vector machine[J]. Journal of Cen-tral South University，2012，43(3): 932-938 (in Chinese)
[2]	Schlkopf B，Smola A，Müller K R. Nonlinear component analysis as a kernel eigenvalue problem[J]. Neural Computation，1998，10(5): 1299-1319
[3]	梁霖，徐光华，栗茂林，等．冲击故障特征提取的非线性流形学习方法[J]. 西安交通大学学报，2009，43(11): 95-99 Liang L，Xu G H，Li M L，et al. Nonlinear manifold learning method of mechanical impact faults extraction[J]. Journal of Xi'an Jiaotong University，2009，43(11): 95-99 (in Chinese)
[4]	Yang W K，Sun C Y，Zhang L. A multi-manifold dis-criminat analysis for image feature extraction[J]. Pattern Recognition，2011，44(8): 1649-1657
[5]	王广斌，李学军，黄良沛．基于 Schur 正交的局部Fisher 判别转子故障诊断[J]. 机械科学与技术，2011，30(1): 62-65 Wang G B，Li X J，Huang L P. Rotor fault diagnosis based on schur orthogonal local fisher discriminant[J]. Mechanical Science and Technology for Aerospace Engi-neering，2011，30(1): 62-65 (in Chinese)
[6]	杜海顺，柴秀丽，汪凤泉，等．一种邻域保持判别嵌入人脸识别方法[J]. 仪器仪表学报，2010，31(3): 625-629Du H S，Chai X L，Wang F Q，et al. Face recognition method using neighborhood preserving discriminant em-bedding[J]. Chinese Journal of Scientific Instrument，2010，31(3): 625-629 (in Chinese)
[7]	王伟，毕笃彦，熊磊，等．基于稀疏流形学习的SAR图像识别[J]. 电子学报，2010，38(11): 2540-2544 Wang W，Bi D Y，Xiong L，et al. SAR image recogni-tion based on sparse manifold learning[J]. Acta Elec-tronica Sinica，2010，38(11): 2540-2544 (in Chinese)
[8]	Cai D，He X F，Zhou K，et al. Locally sensitive discrimi-nant analysis[C]//Proceedings of the 20th International Joint Conference on Artificial Intelligence. California:AAAI press，2007: 708-713
[9]	Roweis S T，Saul L K. Nonlinear dimensionality reduction by locally linear embedding[J]. Science，2000，290 (5500): 2323-2326
[10]	张伟，周维佳，李斌．基于分维 LLE 和 Fisher 判别的故障诊断方法[J]. 仪器仪表学报，2010，31(2):325-333 Zhang W，Zhou W J，Li B. Fault diagnosis approach based on fractal dimension LLE and fisher discriminant[J]. Chinese Journal of Scientific Instrument，2010，31(2): 325-333 (in Chinese)
[11]	李志雄，严新平．独立分量分析和流形学习在VSC-HVDC系统故障诊断中的应用[J]. 西安交通大学学报，2011，45(2): 44-58 Li Z X，Yan X P. Independent component analysis and manifold learning with applications to fault diagnosis of VSC-HVDC systems[J]. Journal of Xi'an Jiaotong Uni-versity，2011，45(2): 44-58 (in Chinese)
[12]	Dijkstra E W. A note on two problems in connexion with graphs[J]. Numerische Mathematik，1959，1 (1): 269-271
[13]	文贵华，江丽君，文军．局部测地距离估计的 Hessian局部线性嵌入[J]. 智能系统学报，2008，3 (5): 429-435 Wen G H，Jiang L J，Wen J. Using locally estimated geodestic distances to improve hessian local linear em-bedding[J]. CAAI Transactions on Intelligent Systems，2008，3(5): 429-435 (in Chinese)
[14]	Lei Y G，He Z J，Zi Y Y. A new approach to intelligent fault diagnosis of rotating machinery[J]. Expert Systems with Applitions，2008，35(4): 1593-1600
[15]	Kouropteva O，Okun O，Hadid A，et al. Beyond locally linear embedding algorithm[R]. Technical Report MVG-01-2002. Machine Vision Group，Uni-versity of Oulu，2002