基于改进DBN的回转支承寿命状态识别

摘要
图/表
参考文献(21)
相关文章 (14)

全文: PDF (1779 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要为了解决大型回转支承背景噪声大，特征信号微弱，寿命状态难以识别等问题，提出了一种基于改进深度信念网络（Deep Belief Network, DBN）的回转支承寿命状态识别方法。DBN网络拥有强大的深度学习能力，能够有效挖掘回转支承运行状态信息，解决了传统浅层网络过度依赖特征提取效果和识别精度不高的问题。在DBN学习训练中，采用新的优化学习方法FEPCD（Free Energy in Persistent Contrastive Divergence），解决了DBN在长期学习中近似和分类能力下降的问题。然后利用自主研发试验台的试验数据对所提方法的优越性进行验证。将改进的DBN算法与浅层分类算法的识别结果进行比较。结果表明改进DBN网络比原始DBN网络和浅层算法能更精确反映回转支承寿命特征，所提方法具有稳定性和智能性的特点。

	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
	王赛赛1
	陈捷1
	2
	王华1
	2
	潘裕斌1

关键词 ：回转支承, 深度学习, 改进DBN, 寿命状态识别

Abstract：In order to solve the problems of large slewing bearing, such as large background noise, weak characteristic signal, and difficulty to identify the life state of slewing bearing, an identification method of slewing bearing life state based on improved deep belief network (DBN) was proposed. DBN network has strong deep learning ability and can effectively mine the running state information of slewing bearing, which solves the problem that traditional shallow network relies too much on the effect of feature extraction and low recognition accuracy. In DBN learning and training, a new optimized learning method,Free Energy in Persistent Contrastive Divergence(FEPCD), was proposed to solve the problem of the decline of approximation and classification ability of DBN in long-term learning. Then the superiority of the proposed method was verified by using the test data of test rig. Finally, the recognition results of the improved DBN algorithm and the shallow classification algorithm were compared. The results show that the improved DBN network can reflect the life characteristics of slewing bearing more accurately than the original DBN network and shallow algorithm, and the proposed method has the characteristics of stability and intelligence.

Key words： Slewing bearing Deep learning Improved DBN Life state recognition

收稿日期: 2018-12-20 出版日期: 2020-03-28

引用本文:

王赛赛1，陈捷1,2，王华1,2，潘裕斌1. 基于改进DBN的回转支承寿命状态识别[J]. 振动与冲击, 2020, 39(7): 238-244.
WANG Saisai1, CHEN Jie1,2, WANG Hua1,2, PAN Yubin1. Life state recognition of slewing bearing based on improved deep belief network. JOURNAL OF VIBRATION AND SHOCK, 2020, 39(7): 238-244.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2020/V39/I7/238

[1] WangF, LiuC, SuW, et al. ConditionmonitoringandfaultdiagnosismethodsforLow-SpeedandHeavy-Loadslewingbearings: Aliteraturereview[J]. Journal ofVibroengineering. 2017, 19(5):3429-3444.

[2] LeeJ, WuF, ZhaoW, etal. Prognosticsandhealthmanagementdesignforrotarymachinerysystems—Reviews, methodologyandapplications[J]. MechanicalSystems & SignalProcessing, 2014, 42(1-2):314-334.

[3]RenL, CuiJ, SunY, et al. Multi-bearingremainingusefullifecollaborativeprediction: Adeeplearningapproach[J]. JournalofManufacturingSystems. 2017, 43:248-256.

[4]Žvokelj, Matej, ZupanS, et al.EEMD-basedmultiscaleICAmethodforslewingbearingfaultdetectionanddiagnosis[J]. JournalofSound & Vibration, 2016, 370:394-423.

[5] LuC, ChenJ, HongR, et al. DegradationtrendestimationofslewingbearingbasedonLSSVMmodel[J]. MechanicalSystemsandSignalProcessing,2016, 76-77: 353-366.

[6] 钮满志, 陈捷, 封杨,等. 基于支持向量机的风电偏航回转支承故障诊断[J]. 南京工业大学学报(自然科学版), 2014, 36(01):117-122.

NIU Man-zhi, CHEN Jie, FENG Yang, et al. Fault diagnosis of wind power yaw slewing bearing based on Support Vector Machine[J]. Journal of Nanjing Tech University (Natural Science Edition), 2014, 36(01): 117-122.

[7] Wang S, Na Z, Lei W, et al. Wind speed forecasting based on the hybrid ensemble empirical mode decomposition and GA-BP neural network method[J]. Renewable Energy, 2016, 94:629-636.

[8] Asr M Y, Ettefagh M M, Hassannejad R, et al. Diagnosis of combined faults in Rotary Machinery by Non-Naive Bayesian approach[J]. Mechanical Systems & Signal Processing, 2017, 85:56-70.

[9] Hinton G E, Salakhutdinov R R. Reducing. the dimensionality of data with neural networks[J], Science, 2006, 313: 504–507.

[10] Bengio Y, Lamblin P, Popovici D, Montreal U. Greedy layer-wise training of deep networks[J]. Advances in Neural Information Processing Systems, 2007, 19: 153-160.

[11]Chen K, Salman A. Learning Speaker-Specific Characteristics with a Deep Neural Architecture[J]. IEEE Transactions on Neural Networks, 2011, 22(11): 1744-1756.

[12] Kuremoto T, Kimura S, Kobayashi K, Obayashi M. Time series forecasting using a deep belief network with restricted Boltzmann machines[J]. Neurocomputing, 2014, 137(15):47-56.

[13] 李巍华, 单外平, 曾雪琼. 基于深度信念网络的轴承故障分类识别[J]. 振动工程学报, 2016(02):340-347.

LI Wei-hua, SAN Wai-ping, CAO Xue-qiong.Fault classification and recognition of bearing based on Deep Belief Network[J].Journal of Vibration Engineering, 2016(02):340-347.

[14] Shao H, Jiang H, Zhang H, et al. Rolling bearing fault feature learning using improved convolutional deep belief network with compressed sensing[J]. Mechanical Systems & Signal Processing, 2018, 100:743-765.

[15] 周晓莉, 张丰, 杜震洪, 等. 基于CRBM算法的时间序列预测模型研究[J]. 浙江大学学报（理学版）, 2016, 43(4):442-451.

ZHOU Xiao-li, ZHANG Feng, DU Zheng-hong, et al. Research on time series prediction model based on CRBM algorithm[J]. Journal of Zhejiang University (Science Edition), 2016, 43(4):442-451.

[16] Chen Z, Deng S, Chen X, et al. Deep neural networks-based rolling bearing fault diagnosis[J]. Microelectronics Reliability, 2017, 75: 327-333.

[17] Hinton G E, Osindero S, Teh Y W. A fast learning algorithm for Deep Belief Nets[J]. Neural Computation, 2006, 18(7):1527–1554.

[18] Fischer A, Igel C. An Introduction to Restricted Boltzmann Machines[J]. 2012, 7441:14-36.

[19] Salakhutdinov R, Murray I. On the quantitative analysis of deep belief networks[C]//International Conference on Machine Learning. 2008: 872-879.

[20] Hinton G E.Training Products of Experts by Minimizing Contrastive Divergence[J]. Neural computation, 2002,14(8): 1771-1800.

[21] Hinton G E. ApracticalguidetotrainingrestrictedBoltzmann machines[J]. Momentum, 2012, 9(1): 599-619.