基于全卷积变分自编码网络FCVAE的轴承剩余寿命预测方法

摘要
图/表
参考文献(17)
相关文章 (15)

全文: PDF (1055 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要由于制造工艺、运行环境等的影响，同型号轴承的使用寿命往往存在较大的个体差异性。在轴承剩余寿命预测中，如果从信号中提取的特征的泛化能力不足，将导致模型预测结果稳定性较差。为此，提出一种基于全卷积变分自编码网络（FCVAE）的轴承的剩余寿命预测方法。该方法用全卷积神经网络（FCNN）改进变分自编码器（VAE），在降低网络复杂度的同时强化所提取特征的泛化能力，并利用频域信号作为模型输入，以进一步降低特征学习的难度，同时设计加权平均方法平滑预测结果。通过试验数据集对所提方法进行验证，结果表明：该方法预测结果的平均误差相比于传统支持向量回归(SVR)降低了64%，比卷积神经网络（CNN）降低45.5%，比VAE降低47.5%。

	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
	张继冬
	邹益胜
	蒋雨良
	曾大毅

关键词 ：全卷积变分自编码, 轴承, 特征提取, 剩余寿命预测

Abstract：Due to influences of manufacturing technology and working condition, service lives of the same type bearings often have great individual difference. If the generalization ability of features extracted in signals is insufficient, the stability of bearing residual life prediction results is poorer. Here, a feature extraction method based on a fully convolutional variational auto-encoder (FCVAE) network was proposed for bearing residual life prediction. In this method, the fully convolutional neural network (FCNN) was used to improve the variational auto-encoder (VAE), reduce the network complexity and strengthen the generalization ability of features extracted. Frequency domain signals were taken as model input to further reduce difficulty of feature learning. At the same time, a weighted average method was designed to smooth the predicted results. Multi-condition test data were used to verify the effectiveness of the proposed method. Results showed that compared with traditional support vector regression (SVR), the average error of prediction results with the proposed method reduces 64%; compared to the convolutional neural network (CNN) and VAE, it reduces 45.5% and 47.5%, respectively.

Key words： fully convolutional variational auto-encoder (FCVAE) bearings feature extraction residual life prediction

收稿日期: 2019-06-19 出版日期: 2020-09-28

引用本文:

张继冬，邹益胜，蒋雨良，曾大毅. 基于全卷积变分自编码网络FCVAE的轴承剩余寿命预测方法[J]. 振动与冲击, 2020, 39(19): 13-18.
ZHANG Jidong, ZOU Yisheng, JIANG Yuliang, ZENG Dayi. Prediction method for bearing residual life based on a FCVAE network. JOURNAL OF VIBRATION AND SHOCK, 2020, 39(19): 13-18.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2020/V39/I19/13

[1] Lei Y , Li N , Guo L , et al. Machinery health prognostics: A systematic review from data acquisition to RUL prediction[J]. Mechanical Systems and Signal Processing, 2018, 104:799-834.
[2] Li N , Lei Y , Lin J , et al. An improved exponential model for predicting remaining useful life of rolling element bearings[J]. IEEE Transactions on Industrial Electronics, 2015, 62(12):1-1.
[3] 李洪儒,于贺,田再克, 等. 基于二元多尺度熵的滚动轴承退化趋势预测[J]. 中国机械工程, 2017, 28(20):2420-2425+2433.
LI Hongru, YU He, TIAN Zaike, et al. Degradation trend prediction of rolling bearing based on two-element multiscale entropy[J]. China Mechanical Engineering, 2017,28(20):2420-2425+2433.
[4] 佘道明,贾民平,张菀.一种新型深度自编码网络的滚动轴承健康评估方法[J].东南大学学报(自然科学版),2018,48(5):801-806.
SHE Daoming, JIA Minping, ZHANG Wan. Deep auto-encoder network method for health assessment of rolling bearings[J]. Journal of Southeast University ( Natural Science Edition ), 2018,48(5): 801-806.
[5] 赵光权,刘小勇,姜泽东,等.基于深度学习的轴承健康因子无监督构建方法[J].仪器仪表学报,2018,39(6):82-88.
ZHAO Guangquan, LIU Xiaoyong, JIANG Zedong, et al. Unsupervised health indicator of bearing based on deep learning. Chinese Journal of Scientific Instrument, 2018, 39(6): 28-88.
[6] Cheng F , Qu L , Qiao W , et al. Enhanced Particle Filtering for Bearing Remaining Useful Life Prediction of Wind Turbine Drivetrain Gearboxes[J]. IEEE Transactions on Industrial Electronics, 2018, 66(6):4738-4748.
[7] 文娟, 高宏力. 一种基于UPF的轴承剩余寿命预测方法[J]. 振动与冲击, 2018, 37(24): 225-230+260.
WEN Juan, GAO Hongli. Remaining useful life prediction of bearings with the unscented particle filter approach[J]. Journal of Vibration and Shock, 2018, 37(24):225-230+260.
[8] 陈法法, 杨勇, 陈保家, 等. 基于模糊信息粒化与小波支持向量机的滚动轴承性能退化趋势预测[J]. 中国机械工程, 2016, 27(12): 1655-1661.
CHEN Fafa, YANG Yong, CHEN Baojia, et al. Degradation trend prediction of rolling bearing based on fuzzy information granulation and wavelet support vector machine[J]. China Mechanical Engineering, 2016, 27(12):1655-1661.
[9] 张星辉, 康建设, 赵劲松, 等. 基于混合高斯输出贝叶斯信念网络模型的设备退化状态识别与剩余使用寿命预测方法研究[J]. 振动与冲击, 2014, 33(8):171-179.
ZHANG Xinghui, KANG Jianshe, ZHAO Jinsong, et al. Equipment degradation state identification and residual life prediction based on MoG-BBN[J]. Journal of Vibration and Shock, 2014, 33(8):171-179.
[10] Jun Z, Nan C, Weiwen P. Estimation of bearing remaining useful life based on multiscale convolutional neural network[J]. IEEE Transactions on Industrial Electronics, 2018, 18(2):466-485.
[11] Li X, Jiang H, Xiong X, et al. Rolling bearing health prognosis using a modified health index based hierarchical gated recurrent unit network[J]. Mechanism and Machine Theory, 2018, 133: 229-249.
[12] 伊恩•古德费洛等著. 深度学习[M]. 北京，人民邮电出版社, 2017: 425-427.
Ian Goodfellow,etc. Deep Learning[M]. Beijing, Posts & Telecom Press, 2017: 425-427.
[13] Kingma D P, Welling M. Auto-Encoding Variational Bayes[C]// International Conference on Learning Representations (ICLR), Banff. 2014
[14] Nectoux P, Gouriveau R, Medjaher K, et al. PRONOSTIA: An experimental platform for bearings accelerated degradation tests[C]// IEEE International Conference on Prognostics and Health Management. IEEE, Denver, 2012: 1-8.
[15] 雷亚国. 旋转机械智能故障诊断与剩余寿命预测[M]. 西安: 西安交通大学出版社, 2017: 294-299.
LEI Yaguo. Intelligent fault diagnosis and remaining useful life prediction of rotating machinery[M]. Xi'an: Xi'an Jiaotong University Press, 2017: 294-299.
[16] Guo L, Lei Y, Li N, et al. Deep convolution feature learning for health indicator construction of bearings[C]// Prognostics and System Health Management Conference. IEEE, Harbin, 2017: 1-6.
[17] 任浩,屈剑锋,柴毅,等. 深度学习在故障诊断领域中的研究现状与挑战[J].控制与决策,2017,32(08):1345-1358.
REN Hao, QU Jianfeng, CHAI Yi, et al. Deep learning for fault diagnosis: The state of the art an challenge[J]. Control and Decision,2017,32(08):1345-1358.