2.793

2018影响因子

(CJCR)

  • 中文核心
  • EI
  • 中国科技核心
  • Scopus
  • CSCD
  • 英国科学文摘

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于无监督深度模型迁移的滚动轴承寿命预测方法

康守强 邢颖怡 王玉静 王庆岩 谢金宝 MIKULOVICHV.I.

康守强, 邢颖怡, 王玉静, 王庆岩, 谢金宝, MIKULOVICHV.I.. 基于无监督深度模型迁移的滚动轴承寿命预测方法. 自动化学报, 2021, x(x): 1−12 doi: 10.16383/j.aas.c200890
引用本文: 康守强, 邢颖怡, 王玉静, 王庆岩, 谢金宝, MIKULOVICHV.I.. 基于无监督深度模型迁移的滚动轴承寿命预测方法. 自动化学报, 2021, x(x): 1−12 doi: 10.16383/j.aas.c200890
Kang Shou-Qiang, Xing Ying-Yi, Wang Yu-Jing, Wang Qing-Yan, Xie Jin-Bao, Mikulovich Vladimir. Rolling bearing life prediction based on unsupervised deep model transfer. Acta Automatica Sinica, 2021, x(x): 1−12 doi: 10.16383/j.aas.c200890
Citation: Kang Shou-Qiang, Xing Ying-Yi, Wang Yu-Jing, Wang Qing-Yan, Xie Jin-Bao, Mikulovich Vladimir. Rolling bearing life prediction based on unsupervised deep model transfer. Acta Automatica Sinica, 2021, x(x): 1−12 doi: 10.16383/j.aas.c200890

基于无监督深度模型迁移的滚动轴承寿命预测方法

doi: 10.16383/j.aas.c200890
基金项目: 国家自然科学基金(51805120); 黑龙江省自然科学基金(LH2019E058); 黑龙江省本科高校青年创新人才培养计划(UNPYSCT-2017091)
详细信息
    作者简介:

    康守强:哈尔滨理工大学电气与电子工程学院教授. 2011 年获得白俄罗斯国立大学博士学位. 主要研究方向为非平稳信号处理, 故障诊断、状态评估与预测技术, 模式识别. E-mail: kangshouqiang@163.com

    邢颖怡:哈尔滨理工大学电气与电子工程学院硕士研究生. 主要研究方向为振动信号处理. E-mail: whale_x@sina.com

    王玉静:哈尔滨理工大学电气与电子工程学院副教授. 2015 年获哈尔滨工业大学博士学位. 主要研究方向为非平稳信号处理, 故障诊断、状态评估与预测技术, 模式识别. 本文通信作者. E-mail: mirrorwyj@163.com

    王庆岩:哈尔滨理工大学电气与电子工程学院讲师. 2018年获哈尔滨工业大学博士学位. 主要研究方向为信号处理、遥感图像智能解译、模式识别. E-mail: wangqy@hrbust.edu.cn

    谢金宝:哈尔滨理工大学电气与电子工程学院副教授. 2012 年获得白俄罗斯国立大学博士学位. 主要研究方向为计算机视觉和自然语言处理. E-mail: jbxpost@163.com

    MIKULOVICHV.I.:Mikulovich Vladimir Ivanovich 白俄罗斯国立大学教授. 1975 年获白俄罗斯国立大学博士学位. 主要研究方向为非平稳信号处理, 故障诊断、状态评估与预测技术, 模式识别. E-mail: falcon@tut.by

Rolling Bearing Life Prediction Based on Unsupervised Deep Model Transfer

Funds: National Natural Science Foundation of China (51805120); Natural Science Foundation of Heilongjiang Province (LH2019E058); University Nursing Program for Young Scholars with Creative Talents in Heilongjiang Province (UNPYSCT-2017091)
More Information
    Author Bio:

    KANG Shou-Qiang Professor at the College of Electrical and Electronic Engineering, Harbin University of Science and Technology, Harbin, China. He received his Ph. D. from Belarusian State University, Minsk, Belarus, in 2011. His research interests include non-stationary signal processing, fault diagnosis, state assessment and prediction technology, pattern recognition. Corresponding author of this article

    XING Ying-Yi Master student at the College of Electrical and Electronic Engineering, Harbin University of Science and Technology. Her research interest is vibration signal processing

    WANG Yu-Jing Associate professor at the College of Electrical and Electronic Engineering, Harbin University of Science and Technology, Harbin, China. She received her Ph. D. from Harbin Institute of Technology, Harbin, China, in 2015. Her research interests include non-stationary signal processing, fault diagnosis, state assessment and prediction technology, pattern recognition

    WANG Qing-Yan Lecturer in the School of Electrical and Electronic Engineering, Harbin University of Science and Technology. He received his Ph.D. from Harbin Institute of Technology in 2018. His research interests include signal processing, intelligent interpretation of remote sensing images, and pattern recognition

    XIE Jin-Bao Associate professor at the College of Electrical and Electronic Engineering, Harbin University of Science and Technology, Harbin, China. He received his Ph. D. from Belarusian State University, Minsk, Belarus, in 2012. His research interests include computer vision and natural language processing

    Mikulovich Vladimir Ivanovich Professor of Belarusian State University, Minsk, Belarus. He received his Ph.D. from Belarusian State University, Minsk, Belarus, in 1975. His research interests include non-stationary signal processing, fault diagnosis, state assessment and prediction technology, pattern recognition

  • 摘要: 针对实际中某种工况滚动轴承带标签振动数据获取困难, 健康指标难以构建及寿命预测误差大的问题, 提出一种基于无监督深度模型迁移的滚动轴承剩余使用寿命预测方法. 该方法首先对滚动轴承全寿命周期振动数据提取均方根特征, 并引入新的自下而上时间序列分割算法将特征序列分割为正常期、退化期和衰退期3种状态; 对振动信号经快速傅里叶变换后的幅值序列进行状态信息标记, 并将其输入到新增卷积层的全卷积神经网络中, 提取深层特征, 得到预训练模型; 提出将预训练模型的梯度作为一种“特征”与传统预训练模型特征一起参与目标域网络训练过程, 从而得到状态识别模型; 利用状态概率估计法结合状态识别模型建立滚动轴承寿命预测模型. 实验验证, 所提方法无需构建健康指标, 可实现无监督条件下不同工况滚动轴承剩余寿命预测, 并获得较好的效果.
  • 图  1  FCN网络结构

    Fig.  1  FCN network structure

    图  2  新增卷积层的FCN网络

    Fig.  2  FCN network with new convolutional layer

    图  3  无监督模型迁移过程示意图

    Fig.  3  Schematic diagram of unsupervised model transfer process

    图  4  logits层训练框图

    Fig.  4  Block diagram of logits layer training

    图  5  滚动轴承剩余寿命预测流程框图

    Fig.  5  Block diagram of remaining life prediction process of the rolling bearing

    图  6  轴承1_1原始数据时域信号

    Fig.  6  Time domain raw signal of bearing 1_1

    图  7  轴承1_1频域幅值信号

    Fig.  7  Frequency domain amplitude signal of bearing 1_1

    图  8  轴承1_1RMS特征的分割结果

    Fig.  8  Segmentation results of bearing 1_1RMS features

    图  9  轴承2_7迁移之前训练及测试损失值

    Fig.  9  Training and testing loss values for bearing 2_7 before transferring

    图  10  轴承2_7迁移之后训练及测试损失值

    Fig.  10  Training and test loss values for bearing 2_7 after transferring

    图  11  所提方法与其他方法的对比结果

    Fig.  11  Comparison of the proposed method with other methods

    表  1  PHM2012数据描述

    Table  1  PHM2012 data description

    数据工况1工况2工况3
    训练数据 轴承1_1 轴承2_1 轴承3_1
    轴承1_2 轴承2_2 轴承3_2
    测试数据 轴承1_3 轴承2_3 轴承3_3
    轴承1_4 轴承2_4
    轴承1_5 轴承2_5
    轴承1_6 轴承2_6
    轴承1_7 轴承2_7
    下载: 导出CSV

    表  2  三种工况描述

    Table  2  Description of the three working conditions

    工况转速/rmp载荷/N
    工况118004000
    工况216504200
    工况315005000
    下载: 导出CSV

    表  3  卷积层数修改前后实验结果

    Table  3  Experimental results before and after modification of the number of convolutional layers

    层数准确率/%
    原始卷积层数目394.23
    修改后卷积数目499.46
    下载: 导出CSV

    表  4  传递梯度特征前后实验对比结果

    Table  4  Experimental comparison results before and after transferring gradient features

    测试集平均准确率/%
    不传递梯度特征传递梯度特征
    1_393.9599.86
    1_499.2999.91
    1_590.9299.91
    1_692.8799.56
    1_787.9098.40
    2_392.8499.92
    2_494.6099.67
    2_591.3598.46
    2_693.7099.57
    2_788.2699.46
    3_392.0499.01
    下载: 导出CSV

    表  5  不同轴承RUL预测误差结果对比

    Table  5  Comparison of RUL prediction error results of different bearings

    不同轴承当前时间点实际预测点本文预测点本文误差EJ(%)
    1_3180105730427025.35
    1_4113802900221023.53
    1_52301016103630−126.87
    1_62301014601540−5.47
    1_7150107570293061.24
    2_3120107530295060.77
    2_461101390118014.49
    2_520010309035088.63
    2_657101290111013.28
    2_7171058033042.11
    3_3351082068016.05
    平均误差19.37
    平均得分0.332
    下载: 导出CSV

    表  6  与其他方法预测误差结果对比

    Table  6  Comparison of prediction error results with other methods

    不同轴承预测误差EJ(%)
    本文方法方案一方案二文献[25]文献[26]
    1_325.3535.9832.5443.28−31.76
    1_423.5340.2536.9267.5562.76
    1_5−126.87−138.54−129.35−22.98−136.03
    1_6−5.47−20.18−15.1821.23−32.88
    1_761.2479.6575.6417.83−11.09
    2_360.7780.2572.4937.8444.22
    2_414.4930.2425.83−19.42−55.40
    2_588.63100.2595.9554.3768.61
    2_613.2835.6830.39−13.95−51.94
    2_742.1160.2155.16−55.17−68.97
    3_316.0540.8135.973.66−21.96
    平均误差19.3731.3328.7632.4853.24
    下载: 导出CSV
  • [1] 毛文涛, 田思雨, 窦智, 张迪, 丁玲. 一种基于深度迁移学习的滚动轴承早期故障在线检测方法. 自动化学报: 1-13[2021-03-22]. https://doi.org/10.16383/j.aas.c190593.

    Mao Wen-Tao, Tian Si-Yu, Dou Zhi, Zhang Di, Ding Ling. An online detection method for early faults of rolling bearings based on deep transfer learning. Acta Automatica Sinica: 1-13[2021-03-22]. https://doi.org/10.16383/j.aas.c190593.
    [2] 张建勋, 杜党波, 司小胜, 胡昌华, 郑建飞. 基于最后逃逸时间的随机退化设备寿命预测方法. 自动化学报: 1-13[2021-03-22]. https://doi.org/10.16383/j.aas.c200260.

    Zhang Jian-Xun, Du Dang-Bo, Si Xiao-Sheng, Hu Chang-Hua, Zheng Jian-Fei. Life prediction method for stochastic degraded equipment based on last escape time. Acta Automatica Sinica: 1-13 [2021-03-22]. https://doi.org/10.16383/j.aas.c200260.
    [3] 康守强, 周月, 王玉静, 谢金宝, V. I. MIKULOVICH. 基于改进SAE和双向LSTM的滚动轴承RUL预测方法. 自动化学报: 1-11[2021-03-22]. https://doi.org/10.16383/j.aas.c190796.

    Kang Shou-Qiang, Zhou Yue, Wang Yu-Jing, Xie Jin-Bao, V. I. MIKULOVICH. Rolling bearing RUL prediction method based on improved SAE and two-way LSTM. Acta Automatica Sinica: 1-11[2021-03-22]. https://doi.org/10.16383/j.aas.c190796.
    [4] 文娟, 高宏力. 一种基于UPF的轴承剩余寿命预测方法. 振动与冲击, 2018, 37(24): 208−213+243

    Wen Juan, Gao Hong-Li. A method for predicting the remaining life of bearings based on UPF. Vibration and Shock, 2018, 37(24): 208−213+243
    [5] 裴洪, 胡昌华, 司小胜, 张建勋, 庞哲楠, 张鹏. 基于机器学习的设备剩余寿命预测方法综述. 机械工程学报, 2019, 55(8): 1−13 doi: 10.3901/JME.2019.08.001

    Pei Hong, Hu Chang-Hua, Si Xiao-Sheng, Zhang Jian-Xun, Pang Zhe-Nan, Zhang Peng. Overview of equipment remaining life prediction methods based on machine learning. Journal of Mechanical Engineering, 2019, 55(8): 1−13 doi: 10.3901/JME.2019.08.001
    [6] 雷亚国, 李乃鹏, 林京. 基于粒子滤波的滚动轴承寿命预测方法. 中国机械工程学会可靠性工程分会. 2014年全国机械行业可靠性技术学术交流会暨可靠性工程分会第五届委员会成立大会论文集. 中国机械工程学会可靠性工程分会: 中国机械工程学会, 2014.198−203

    Lei Ya-Guo, Li Nai-Peng, Lin Jing. Rolling bearing life prediction method based on particle filter. Reliability Engineering Branch of Chinese Mechanical Engineering Society. 2014 National Reliability Technology Academic Exchange Conference and Reliability Engineering Branch of National Machinery Industry Proceedings of the Founding Conference of the 5th Committee. Reliability Engineering Branch of Chinese Mechanical Engineering Society: Chinese Mechanical Engineering Society, 2014.198−203
    [7] 杨宇, 张娜, 程军圣. 全参数动态学习深度信念网络在滚动轴承寿命预测中的应用. 振动与冲击, 2019, 38(10): 199−205+249

    Yang Yu, Zhang Na, Cheng Jun-Sheng. Application of full-parameter dynamic learning deep belief network in rolling bearing life prediction. Vibration and Shock, 2019, 38(10): 199−205+249
    [8] 瞿家明, 周易文, 王恒, 黄希, 姜杰. 基于改进HMM和Pearson相似度分析的滚动轴承自适应寿命预测方法. 振动与冲击, 2020, 39(8): 172−177+201

    Qu Jia-Ming, Zhou Yi-Wen, Wang Heng, Huang Xi, Jiang Jie. Adaptive life prediction method for rolling bearings based on improved HMM and Pearson similarity analysis. Vibration and Shock, 2020, 39(8): 172−177+201
    [9] 王玉静, 王诗达, 康守强, 王庆岩, V. I. MIKULOVICH. 基于改进深度森林的滚动轴承剩余寿命预测方法. 中国电机工程学报, 2020, 40(15): 5032−5043

    Wang Yujing, Wang Shida, Kang Shouqiang, Wang Qing-Yan, V. I. MIKULOVICH. Remaining life prediction method of rolling bearing based on improved deep forest. Proceedings of the Chinese Society of Electrical Engineering, 2020, 40(15): 5032−5043
    [10] Guo T, Deng Z. An improved EMD method based on the multi-objective optimization and its application to fault feature extraction of rolling bearing. Applied Acoustics, 2017: 46−62
    [11] Xu Y, Cao J, Zhao J. Application of fast singular spectrum decomposition method based on order statistic filter in rolling bearing fault diagnosis. Measurement Science and Technology, 2019, 30(12
    [12] 赵春华, 胡恒星, 陈保家, 张毅娜, 肖嘉伟. 基于深度学习特征提取和WOA-SVM状态识别的轴承故障诊断. 振动与冲击, 2019, 38(10): 31−37+48

    Zhao Chun-hua, Hu Heng-Xing, Chen Bao-Jia, Zhang Yi-Na, Xiao Jia-Wei. Bearing fault diagnosis based on deep learning feature extraction and WOA-SVM state recognition. Journal of Vibration and Shock, 2019, 38(10): 31−37+48
    [13] Xu F, Tse P W. Automatic roller bearings fault diagnosis using DSAE in deep learning and CFS algorithm. Soft computing, 2019, 23(13): 5117−5128 doi: 10.1007/s00500-018-3178-x
    [14] 王玉静, 那晓栋, 康守强, 谢金宝, V I MIKULOVICH. 基于EEMD-Hilbert包络谱和DBN的变负载下滚动轴承状态识别方法. 中国电机工程学报, 2017, 37(23): 6943−6950+7085

    Wang Yu-Jing, Na Xiao-Dong, Kang Shou-Qiang, Xie Jin-Bao, V. I. MIKULOVICH. State identification method of rolling bearing under variable load based on EEMD-Hilbert envelope spectrum and DBN. Chinese Journal of Electrical Engineering, 2017, 37(23): 6943−6950+7085
    [15] 昝涛, 王辉, 刘智豪, 王民, 高相胜. 基于多输入层卷积神经网络的滚动轴承故障诊断模型. 振动与冲击, 2020, 39(12): 142−149+163

    Zan Tao, Wang Hui, Liu Zhi-Hao, Wang Min, Gao Xiang-Sheng. Fault diagnosis model of rolling bearing based on multi-input layer convolutional neural network. Vibration and Shock, 2020, 39(12): 142−149+163
    [16] Guo L, Lei Y, Xing S. Deep Convolutional Transfer Learning Network: A New Method for Intelligent Fault Diagnosis of Machines With Unlabeled Data. IEEE Transactions on Industrial Electronics, 2019, 66(9): 7316−7325 doi: 10.1109/TIE.2018.2877090
    [17] Li J, Li X, He D. A domain adaptation model for early gear pitting fault diagnosis based on deep transfer learning network. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, 2019: 1748006X19867776
    [18] Zhang R, Tao H, Wu L, Guan Y. Transfer Learning With Neural Networks for Bearing Fault Diagnosis in Changing Working Conditions. IEEE Access, 2017: 1−1
    [19] Eker O F, Camci F. State-based prognostics with state duration information. Quality and Reliability Engineering International, 2013, 29(4): 465−476 doi: 10.1002/qre.1393
    [20] Lin J, Keogh E, Truppel W. Clustering of streaming time series is meaningless. Proceedings of the 8th ACM SIGMOD workshop on Research issues in data mining and knowledge discovery. 2003. 56−65
    [21] Atamuradov V, Medjaher K, Camci F. Feature selection and fault-severity classification-based machine health assessment methodology for point machine sliding-chair degradation. Quality and Reliability Engineering International, 2019, 35(4): 1081−1099 doi: 10.1002/qre.2446
    [22] Bui D T, Hoang N D, Martínez-Álvarez F. A novel deep learning neural network approach for predicting flash flood susceptibility: A case study at a high frequency tropical storm area. Science of The Total Environment, 2020, 701: 134413 doi: 10.1016/j.scitotenv.2019.134413
    [23] 齐申武. 基于振动信号分析的滚动轴承故障诊断及寿命预测研究. 燕山大学, 2018

    Qi Shen-Wu. Research on rolling bearing fault diagnosis and life prediction based on vibration signal analysis. Yanshan University, 2018
    [24] PATRICK N, RAFAEL G, KAMAL M. PRONOSTIA: Anexperimental platform for bearing saccelerated life test. IEEE International Conference on Prognostics and Health Management. USA: Denver, 2012.1−8
    [25] Guo L, Li N P, Jia F, Lei Y G, Lin J. A recurrent neural network based health indicator for remaining useful life prediction of bearings. Neurocomputing, 2017, 240: 98−1093 doi: 10.1016/j.neucom.2017.02.045
    [26] Hong S, Zhou Z, Zio E, Hong K. Condition assessment for the performance degradation of bearing based on a combinatorial feature extraction method. Digital Signal Processing, 2014, 27: 159−166 doi: 10.1016/j.dsp.2013.12.010
  • 加载中
计量
  • 文章访问数:  144
  • HTML全文浏览量:  69
  • 被引次数: 0
出版历程
  • 收稿日期:  2020-10-25
  • 修回日期:  2021-03-19
  • 网络出版日期:  2021-06-07

目录

    /

    返回文章
    返回