针对变工况环境下采集到的滚动轴承寿命状态数据存在特征分布差异，深度神经网络模型泛化能力差的问题，本文结合时间卷积网络（Temporal Convolutional Neural Network，TCN）和残差自注意力机制提出了一种端到端的滚动轴承剩余寿命（Remaining Useful Life，RUL）迁移预测方法。首先，将传感器采集到的一维时域信号利用短时傅里叶变换转换为频域信号；其次，剩余寿命迁移预测网络通用特征提取层采用残差自注意力TCN网络，该网络在较好提取时间序列信息的同时，进一步通过残差自注意力机制捕获轴承局部退化特征，增强模型的迁移特征提取能力；再次，采用提出的联合领域自适应策略匹配变工况下滚动轴承寿命状态数据特征分布差异，实现不同工况下轴承寿命状态知识的迁移预测；最后，在公开的滚动轴承全寿命数据集上进行试验验证，结果表明本文所提方法能有效实现变工况下的滚动轴承RUL预测，并获得较好的预测性能。
Abstract
Aiming at the problem of the difference in characteristic distribution of the collected data of rolling bearing life status under variable working conditions and the poor generalization ability of the deep neural network model, an end-to-end transfer prediction method of rolling bearing remaining life (RUL) is proposed in this paper, combining the temporal convolutional neural network (TCN) and the residual self-attention mechanism (RSAM). Firstly, the one-dimensional time-domain signal collected by the sensor was converted into a frequency-domain signal by short-time Fourier transform; Secondly, the proposed network was used in the general feature extraction layer of the transfer residual life prediction network. While extracting time series information, the RSAM was further used to capture the local degradation features of the bearing, which enhances the model's ability to extract transfer features; Thirdly, the proposed joint domain self-adaptive strategy is used to match the characteristic distribution difference of rolling bearing life state data under variable working conditions, so as to realize the migration prediction of bearing life state information under different working conditions; Finally, the experimental verification was carried out on the full life of the rolling bearing. The results show that the proposed method can effectively realize the RUL prediction of the rolling bearing under variable working conditions, and obtain a better prediction. performance.
关键词
剩余寿命 /
滚动轴承 /
时间卷积网络 /
残差自注意力 /
迁移学习
{{custom_keyword}} /
Key words
remaining useful life /
rolling bearings /
temporal convolutional neural network /
residual self-attention /
transfer learning.
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 裴洪，胡昌华，司小胜，等. 基于机器学习的设备剩余寿命预测方法综述[J]. 机械工程学报，2019，55(08)：1-13.
PEI Hong, HU Changhua, SI Xiaosheng, et al. Overview of machine learning-based equipment remaining life prediction methods[J]. Chinese Journal of Mechanical Engineering, 2019, 55(08): 1-13.
[2] GUO Jian, LI Zhaojun, LI Meiyan. A review on prognostics methods for engineering systems[J]. IEEE Transactions on Reliability, 2020, 69(03): 1110-1129.
[3] CHEN Yuanhang, PENG Gaoliang, ZHU Zhiyu, et al. A novel deep learning method based on attention mechanism for bearing remaining useful life prediction[J]. Applied Soft Computing Journal, 2019, 86: 105919.
[4] 赵志宏，李晴，杨绍普，等. 基于BiLSTM与注意力机制的剩余使用寿命预测研究[J]. 振动与冲击，2022，41(06): 44-50+196.
Zhao Zhihong, Li Qing, Yang Shaopu, et al. Research on residual service life prediction based on BiLSTM and attention mechanism[J]. Vibration and Shock, 2022,41 (06): 44-50+196.
[5] 王玉静，李少鹏，康守强，等. 结合CNN和LSTM的滚动轴承剩余使用寿命预测方法[J].振动.测试与诊断，2021，41(03)：439-446+617.
Wang Yujing, Li Shaopeng, Kang Shouqiang, et al. Remaining service life prediction method of rolling bearing combined with CNN and LSTM[J]. Vibration. Testing and Diagnosis, 2021, 41(03): 439-446+617.
[6] 陈保家，陈正坤，陈学良，等. 基于注意力TCN的滚动轴承剩余使用寿命预测方法[J]. 电子测量技术，2021，44(24): 153-160.
Chen Baojia, Chen Zhengkun, Chen Xueliang, et al. Remaining service life prediction method of rolling bearing based on attention TCN[J]. Electronic Measurement Technology, 2021, 44(24): 153-160.
[7] Wang Y, Deng L, Zheng L, et al. Temporal convolutional network with soft thresholding and attention mechanism for machinery prognostics[J]. Journal of Manufacturing Systems, 2021, 60:512-526.
[8] Zhu J, Chen N, Shen C. A new data-driven transferable remaining useful life prediction approach for bearing under different working conditions[J]. Mechanical Systems and Signal Processing, 2020, 139: 106602.
[9] 朱朋，董绍江，李洋，等. 基于残差注意力机制和子领域自适应的时变转速下滚动轴承故障诊断[J]. 振动与冲击，2022，41(22): 293-300.
Zhu Peng, Dong Shaojiang, Li Yang, et al. Fault diagnosis of rolling bearing under time-varying rotational speed based on residual attention mechanism and sub domain adaptation[J]. Vibration and Shock, 2022, 41 (22): 293-300.
[10] 王新刚，韩凯忠，王超，等. 基于迁移学习的轴承剩余使用寿命预测方法[J].东北大学学报(自然科学版)，2021，42(05): 665-672.
Wang Xingang, Han Kaizhong, Wang Chao, et al. Prediction method of bearing remaining service life based on transfer learning[J]. Journal of Northeastern University (Natural Science Edition), 2021, 42(05): 665-672.
[11] Zeng F, Li Y, Jiang Y, et al. An online transfer learning-based remaining useful life prediction method of ball bearings[J]. Measurement, 2021, 176: 109201.
[12] Cao Y, Ding Y, Jia M, et al. A novel temporal convolutional network with residual self-attention mechanism for remaining useful life prediction of rolling bearings[J]. Reliability Engineering and System Safety, 2021, 215: 107813.
[13] Hu T, Guo Y, Gu L, et al. Remaining useful life estimation of bearings under different working conditions via Wasserstein distance-based weighted domain adaptation[J]. Reliability Engineering and System Safety, 2022, 224: 108526.
[14] GANIN Y, LEMPITSKKY V. Unsupervised domain adaptation by backpropagation[C]//International conference on machine learning. PMLR, 2015: 1180-1189.
[15] Bai S, Kolter J Z, Koltun V. An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling[J]. ArXiv Preprint, 2018, ArXiv: 1803.01271.
[16] 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, 2012.
[17] 姚德臣，李博阳，刘恒畅，等. 基于注意力GRU算法的滚动轴承剩余寿命预测[J]. 振动与冲击，2021，40(17): 116-123.
Yao Dechen, Li Boyang, Liu Hengchang, et al. Residual Life Prediction of Rolling Bearing Based on Attention GRU Algorithm[J]. Vibration and Shock, 2021, 40 (17): 116-123.
[18] SUN B, SAENKO K. Deep coral: Correlation alignment for deep domain adaptation[C]//European conference on computer vision. Springer, Cham, 2016: 443-450.
[19] Xia P, Huang Y, Li P, et al. Fault Knowledge Transfer Assisted Ensemble Method for Remaining Useful Life Prediction[J]. IEEE Transactions on Industrial Informatics, 2022, 18(03): 1758-1769.
.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(1752 KB)
