作者：胡城豪, 胡昌华, 司小胜, 杜党波, 高旭东

作者单位：火箭军工程大学导弹工程学院，陕西 西安 710025

关键词：轴承;CNN;LSTM;剩余寿命预测;多尺度技术

摘要：

针对现有卷积神经网络(convolutional neural network, CNN)与长短时记忆网络(long short-term memory, LSTM)堆叠的寿命预测方法忽略低层次信息的问题，引入多尺度技术，提出一种多尺度卷积长短时记忆网络模型(multi-scale CNN-LSTM, MSCNN-LSTM)。将CNN的输出由单一尺度转换为多尺度，以充分学习CNN模块提取到的不同层次退化特征。首先采用小波变换获取退化信号的时频信息，并根据初始时刻标准差划分健康阶段；而后利用退化阶段监测数据训练所构建的多尺度网络；最后使用该网络预测旋转机械剩余寿命。在PHM 2012轴承数据集上的验证结果表明，所提MSCNN-LSTM模型能够同时学习退化数据中的低层次和高层次信息，有效提高轴承的剩余寿命预测精度。

Remaining useful lifetime prediction method of rolling bearing based on MSCNN-LSTM
HU Chenghao, HU Changhua, SI Xiaosheng, DU Dangbo, GAO Xudong
School of Missile Engineering, Rocket Force University of Engineering, Xi’an 710025, China
Abstract: The existing model consisting of the convolutional neural network (CNN) and the long short-term memory (LSTM) ignores the low level information extracted from CNN, multi-scale technology is introduced to propose a multi-scale CNN-LSTM model (MSCNN-LSTM) which converts the output of CNN from single scale to multi-scale in order to fully learn the degradation characteristics of different levels extracted from CNN. Firstly, the time-frequency information of the signal is obtained using the wavelet transform, and the health stage is divided according to the initial time's standard deviation. Then the monitoring data of degradation phase is used to train the multi-scale network. Finally, the network is used to predict the remaining useful lifetime of bearings. The verification result on PHM 2012 bearing data set shows that the proposed MSCNN-LSTM model can learn both low-level and high-level information from monitoring data at the same time, effectively improving the prediction accuracy of the remaining useful lifetime of bearings.
Keywords: bearing;CNN;LSTM;remaining useful lifetime prediction;multi-scale technology
2020, 46(8):103-110  收稿日期: 2020-05-31;收到修改稿日期: 2020-07-08
基金项目: 国家自然科学基金（61833016,61903376,61773389）
作者简介: 胡城豪（1996-），男，山东烟台市人，硕士研究生，专业方向为深度学习、机械设备剩余寿命预测
参考文献
[1] 任子强, 司小胜, 胡昌华, 等. 融合多源数据的非线性退化建模与剩余寿命预测[J]. 中国测试, 2020, 46(2): 1-8
[2] 张鹏, 胡昌华, 白灿, 等. 考虑随机效应的两阶段退化系统剩余寿命预测方法[J]. 中国测试, 2019, 45(1): 1-7
[3] 田凯. 海底输油管道在悬空与腐蚀联合作用下的剩余寿命预测[J]. 中国测试, 2018, 44(2): 6-10
[4] YIN S, XIE X, SUN W. A nonlinear process monitoring approach with locally weighted learning of available data[J]. IEEE Transactions on Industrial Electronics, 2017, 64(2): 1507-1516
[5] 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
[6] LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553): 436-444
[7] 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(8): 1-13
[8] HU C H, PEI H, SI X S, et al. A prognostic model based on DBN and diffusion process for degrading bearing[J]. IEEE Transactions on Industrial Electronics, 2020, 67(10): 8767-8777
[9] WANG B, LEI Y, YAN T, et al. Recurrent convolutional neural network: a new framework for remaining useful life prediction of machinery[J]. Neurocomputing, 2020, 379: 117-129
[10] MAO W, HE J, TANG J, et al. Predicting remaining useful life of rolling bearings based on deep feature representation and long short-term memory neural network[J]. Advances in Mechanical Engineering, 2018, 10(12): 1-18
[11] 李少鹏. 结合CNN和LSTM的滚动轴承剩余使用寿命预测方法研究[D]. 哈尔滨: 哈尔滨理工大学, 2019.
[12] 曾大懿, 杨基宏, 邹益胜, 等. 基于并行多通道卷积长短时记忆网络PMCCNN-LSTM的轴承寿命预测方法[J/OL]. 中国机械工程: 1-9[2020-05-27].http://kns.cnki.net/kcms/detail/42. 1294.TH.20200629.1544.010.html.
[13] LI X, ZHANG W, DING Q. Deep learning-based remaining useful life estimation of bearings using multi-scale feature extraction[J]. Reliability Engineering and System Safety, 2019, 182: 208-218
[14] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//Communications of the ACM, 2017: 1097-1105.
[15] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324
[16] SUN Y, WANG X, TANG X. Deep learning face representation from predicting 10, 000 classes[J]. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2014, 7(4): 1891-1898
[17] RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088): 696-699
[18] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780
[19] YU Y, SI X, HU C, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Computation, 2019, 31(7): 1235-1270
[20] YU Y, HU C, SI X, et al. Averaged Bi-LSTM networks for RUL prognostics with non-life-cycle labeled dataset[J]. Neurocomputing, 2020
[21] 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, 2012.
[22] SOUALHI A, MEDJAHER K, ZERHOUNI N. Bearing health monitoring based on hilbert-huang transform, support vector machine, and regression[J]. IEEE Transactions on Instrumentation and Measurement, 2015, 64(1): 52-62
[23] SINGLETON R K, STRANGAS E G, AVIYENTE S. Extended kalman filtering for remaining-useful-life estimation of bearings[J]. IEEE Transactions on Industrial Electronics, 2014, 62(3): 1781-1790
[24] GROSSMANN A, MORLET J. Decomposmon of hardy functions into square integrable wavelets of constant shape[J]. Fundamental Papers in Wavelet Theory, 2009, 15(4): 723-736
[25] TANG B, LIU W, SONG T. Wind turbine fault diagnosis based on Morlet wavelet transformation and Wigner-Ville distribution[J]. Renewable Energy, 2010, 35(12): 2862-2866
[26] ZHU J, CHEN N, PENG W. Estimation of bearing remaining useful life based on multiscale convolutional neural network[J]. IEEE Transactions on Industrial Electronics, 2019, 66(4): 3208-3216
[27] RAVEENDRAN H, THOMAS D. Image fusion using LEP filtering and bilinear interpolation[J]. International Journal of Engineering Trends and Technology, 2014, 12(9): 427-431
[28] SAXENA A, CELAYA J, SAHA B. et al Metrics for offline evaluation of prognostic performance[J]. International Journal of Prognostics and Health Management, 2010, 1(1): 2153-2648