改进TrAdaBoost多分类算法的滚动轴承故障诊断

摘要
图/表
参考文献(15)
相关文章 (15)

全文: PDF (863 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要实际工程中滚动轴承受工况、运行环境等因素影响，获取的数据不易满足传统机器学习中训练数据和测试数据独立同分布且训练样本足够多的条件，直接影响故障诊断率。为此，提出一种改进TrAdaBoost多分类算法的滚动轴承故障诊断方法。首先，引入大量辅助标记数据和少量目标标记数据组成联合训练集使训练样本足够多，并应用异分布加权随机抽样对TrAdaBoost迭代过程中的联合训练集进行重组，获得与测试集“近似同分布”的优化联合训练集，降低不同分布数据间的差异性。其次，将迭代结束后的内部分类器模型作为输出，从而改变TrAdaBoost的输出机制使其适应多分类任务。最后，为削弱随机抽样对诊断结果的影响，对多次抽样得到的结果进行一致性投票以得到最终诊断结果。实验结果证明了所提方法的可行性与有效性。

	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
	陈仁祥1
	2
	陈思杨1
	杨黎霞3
	徐向阳1
	董绍江1
	唐林林1

关键词 ：滚动轴承, 故障诊断, 加权随机抽样, TrAdaBoost

Abstract：In engineering practice, rolling bearings are affected by working conditions and runtime environment.The acquired data can’t easily satisfy conditions of training data and tested ones being independent and having the same distribution, and enough training samples in traditional machine learning to directly affect fault diagnosis rate.Here, a rolling bearing fault diagnosis method based on the improved TrAdaBoost multi-classification algorithm was proposed.Firstly, a large amount of auxiliary labeled data and a small amount of target labeled data were introduced to form a joint training set, and make training samples enough.The heterogeneous distribution weighted random sampling was used to reconstruct the joint training set in TrAdaBoost iteration process, acquire the optimal joint training set with the approximate same distribution as that of the tested set,and reduce the diversity among different distribution data.Secondly, the internal classifier model after iteration ending was taken as output to change the output mechanism of TrAdaBoost, and make it adapt to the multi-class task.Finally, to weaken effects of random sampling on diagnosis results, results of multiple samplings were voted with consistence to get the final diagnosis results.The test results verified the feasibility and effectiveness of the proposed method.

Key words： rolling bearings fault diagnosis weighted random sampling TrAdaBoost

收稿日期: 2018-02-27 出版日期: 2019-07-28

引用本文:

陈仁祥1,2，陈思杨1，杨黎霞3，徐向阳1，董绍江1，唐林林1. 改进TrAdaBoost多分类算法的滚动轴承故障诊断[J]. 振动与冲击, 2019, 38(15): 36-41.
CHEN Renxiang1,2, CHEN Siyang1, YANG Lixia3, XU Xiangyang1, DONG Shaojiang1, TANG Linlin1. Fault diagnosis of rolling bearings based on improved TrAdaBoost multi-classification algorithm. JOURNAL OF VIBRATION AND SHOCK, 2019, 38(15): 36-41.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2019/V38/I15/36

[1] 杨宇, 何知义, 潘海洋，等. 基于LCD-Hilbert谱奇异值和QRVPMCD的滚动轴承故障状态识别方法[J]. 振动与冲击, 2015(7):121-126.
YANG Yu，HE Zhi-yi，PAN Hai-yan，et al. The rolling bearing fault diagnosis method based on the Hilbert spectrum singular value and QRVPMCD. Journal of Vibration and Shock, 2015, 34(7): 121-126.
[2] 庄福振，罗平，何清，等.迁移学习研究进展[J].软件学报,2015, 26(1):26-39.
ZHUANG Fu-zhen，LUO Ping，HE Qing，et al. Survey on transfer learning research. Journal of Software, 2015, 26(1):26−39(in Chinese).
[3] 赵孝礼, 赵荣珍, 孙业北,等. 基于正则化核最大边界投影维数约简的滚动轴承故障状态识别[J]. 振动与冲击, 2017, 36(14).
ZHAO Xiao-li，ZHAO Rong-zhen，SUN Ye-bei，et al. Fault Diagnosis of Rolling Bearing Based on Dimension Reduction with Regularized Kernel Maximum Margin Projection. Journal of Vibration and Shock, 2017, 36(14): 104-110.
[4] Adil M, Abid M, Khan A Q, et al. Exponential discriminant analysis for fault diagnosis[J]. Neurocomputing, 2016, 171(C):1344-1353.
[5] 刘若晨, 左洪福. 变工况下滚动轴承故障注入静电监测方法研究[J]. 仪器仪表学报, 2014, 35(10):2348-2355.
LIU Ruo-chen，ZUO Hong-fu. Research on electrostatic monitoring method of rolling bearings with injected fault under variable operating conditions[J]. Chinese Journal of Scientific Instrument, 2014, 35(10):2348-2355.
[6] Borghesani P, Ricci R, Chatterton S, et al. A new procedure for using envelope analysis for rolling element bearing diagnostics in variable operating conditions[J]. Mechanical Systems & Signal Processing, 2013, 38(1):23-35.
[7] Pan S J, Yang Q. A Survey on Transfer Learning[J]. IEEE Transactions on Knowledge & Data Engineering, 2010, 22(10):1345-1359.
[8] Wieschollek P, Lensch H P A. Transfer Learning for Material Classification using Convolutional Networks[J]. 2016.
[9] Rajagopal A K, Subramanian R, Vieriu R L, et al. Exploring Transfer Learning Approaches for Head Pose Classification from Multi-view Surveillance Images[J]. International Journal of Computer Vision, 2014, 109(1-2):146-167.
[10] 陈超,沈飞,严如强.改进LSSVM迁移学习方法的轴承故障状态识别[J].仪器仪表学报, 2017, 38(1):33-40.
CHEN Chao, SHEN Fei, YAN Ru-qiang. Enhanced least squares support vector machine-based transfer learning strategy for bearing fault diagnosis[J]. Chinese Journal of Scientific Instrument, 2017,38(1):33-40.
[11] 段礼祥,谢骏遥,王凯,等.基于不同工况下源领域数据集的齿轮箱故障状态识别[J].振动与冲击, 2017(10):104-108.
DUAN Li-xiang, XIE Jun-yao, WANG Kai, et al . Considering the Influence of Kernel Function of Transfer Component Analysis and its Application in Variable Working Condition of Gearbox Fault Diagnosis. Journal of Vibration and Shock, 2017, 36(10): 104-108.
[12] Dai W, Yang Q, Xue G R, et al. Boosting for transfer learning[C]// International Conference on Machine Learning. ACM, 2007:193-200.
[13] 张倩, 李明, 王雪松,等. 一种面向多源领域的实例迁移学习[J]. 自动化学报, 2014, 40(6):1176-1183.
ZHANG Qian, LI Ming, WANG Xue-Song, CHENG Yu-Hu, ZHU Mei-Qiang. Instance-based Transfer Learning for Multi-source Domains. Acta Automatica Sinica, 2014, 40(6): 1176-1183.
[14] Efraimidis P S. Weighted Random Sampling over Data Streams[J]. 2015, 9295:183-195.
[15] 陈仁祥, 陈思杨, 杨黎霞,等. 基于振动敏感时频特征的航天轴承寿命状态识别方法[J]. 振动与冲击, 2016, 35(17):134-139.
CHEN Ren-xiang, CHEN Si-yang, YANG Lixia, et al. Life state recognition method for space bearings based on sensitive time-frequency feature of vibration. Journal of Vibration and Shock, 2016, 35(17): 134-139.