基于自适应马氏空间与深度学习的滚动轴承退化趋势预测

doi:10.12305/j.issn.1001-506X.2023.10.39

摘要/Abstract

摘要：

滚动轴承退化趋势预测中, 传统特征选择主要依赖人工经验和单一评价算法, 容易造成特征的少选或错选, 且单一深度学习网络无法充分挖掘数据中包含的性能退化信息, 导致模型预测精度较低。针对上述问题, 提出一种基于自适应马氏空间(adaptive Mahalanobis space, AMS)与融合深度学习网络的滚动轴承退化趋势预测方法。首先, 分解原始信号并利用相关峭度系数准则筛选固有模态函数(intrinsic mode function, IMF)分量重构新信号, 从多域视角提取特征; 然后, 构建基于AMS的多目标特征选择算法自动优选特征, 减少人工依赖, 加强自适应性和泛化性, 并将马氏距离(Mahalanobis distance, MD)与指数加权移动平均(exponential weighted moving average, EWMA)方法进行结合，对轴承性能退化趋势进行良好表征；最后, 利用稀疏自动编码器和门控循环单元(sparse auto encoder-gated recurrent unit, SAE-GRU)融合模型进行预测。实验结果表明, 所提方法能够有效筛选最优特征, 显著提高预测精度。

关键词: 滚动轴承, 自适应马氏空间, 多目标特征选择, 融合模型, 退化趋势预测

Abstract:

In the prediction of rolling bearing degradation trend, traditional feature selection mainly relies on manual experience and a single evaluation algorithm, which is easy to cause under-selection or misselection of features. Moreover, a single deep learning network cannot fully mine the performance degradation information contained in the data, which results in low prediction accuracy of the model. To solve the above problem, a prediction method of rolling bearing degradation trend based on adaptive Mahalanobis space (AMS) and fusion deep learning network is proposed. Firstly, the original signals are decomposed and the correlated kurtosis coefficient criterion is used to screen the intrinsic mode function (IMF) to reconstruct the new signals, and the features are extracted from the multi-domain perspective. Secondly, the multi-objective feature selection algorithm based on AMS is built to optimize characteristic automatically. With the aim of reducing manual dependencies, and serengthening the adaptability and generalization, Mahalanobis distance (MD) is combined with the exponential weighted moving average (EWMA) method to well characterize the degradation trend of bearing performance. Finally, the fusion model of sparse auto encoder and gated recurrent unit (SAE-GRU) is used for prediction. The experimental results show that the proposed method can effectively screen the optimal features and significantly improve the prediction accuracy.

Key words: rolling bearing, adaptive Mahalanobis space (AMS), multi-objective feature selection, fusion model, degradation trend prediction

中图分类号:

TH17

吴梦蝶, 程龙生, 陈闻鹤. 基于自适应马氏空间与深度学习的滚动轴承退化趋势预测[J]. 系统工程与电子技术, 2023, 45(10): 3338-3349.

Mengdie WU, Longsheng CHENG, Wenhe CHEN. Degradation trend prediction of rolling bearing based on adaptive Mahalanobis space and deep learning[J]. Systems Engineering and Electronics, 2023, 45(10): 3338-3349.

图/表 24

表1

图1

图2

图3

图4

图5

图6

图7

图8

表2

图9

表3

表4

表5

图10

图11

表6

图12

表7

图13

图14

表8

图15

图16

参考文献 29

1	LI X , ZHANG W , DING Q . Deep learning-based remaining useful life estimation of bearings using multi-scale feature extraction[J]. Reliability Engineering & System Safety, 2019, 182 (2): 208- 218.
2	KUMAR A , PARKASH C , VASHISHTHA G , et al. State-space modeling and novel entropy-based health indicator for dynamic degradation monitoring of rolling element bearing[J]. Reliability Engineering & System Safety, 2022, 221, 108356.
3	KUMAR A , GANDHI C P , VASHISHTHA G , et al. VMD based trigonometric entropy measure: a simple and effective tool for dynamic degradation monitoring of rolling element bearing[J]. Mea-surement Science and Technology, 2021, 33 (1): 014005.
4	HALL M A. Correlation-based feature selection of discrete and numeric class machine learning[C]//Proc. of the 17th International Conference on Machine Learning, 2000: 359-366.
5	YANG F , HABIBULLAH M S , ZHANG T Y , et al. Health index-based prognostics for remaining useful life predictions in electrical machines[J]. IEEE Trans.on Industrial Electronics, 2016, 63 (4): 2633- 2644. doi: 10.1109/TIE.2016.2515054
6	ZHANG B , ZHANG L J , XU J W . Degradation feature selection for remaining useful life prediction of rolling element bearings[J]. Quality and Reliability Engineering International, 2016, 32 (2): 547- 554. doi: 10.1002/qre.1771
7	CHEN J L , JING H J , CHANG Y H , et al. Gated recurrent unit based recurrent neural network for remaining useful life prediction of nonlinear deterioration process[J]. Reliability Engineering & System Safety, 2019, 185, 372- 382.
8	NGUYEN K T P , MEDJAHER K . A new dynamic predictive maintenance framework using deep learning for failure prognostics[J]. Reliability Engineering & System Safety, 2019, 188, 251- 262.
9	AHMAD W , KHAN S A , ISLAM M M M , et al. A reliable technique for remaining useful life estimation of rolling element bearings using dynamic regression models[J]. Reliability Engineering & System Safety, 2019, 184, 67- 76.
10	ZHAO R , YAN R Q , CHEN Z H , et al. Deep learning and its applications to machine health monitoring[J]. Mechanical Systems and Signal Processing, 2019, 115, 213- 237. doi: 10.1016/j.ymssp.2018.05.050
11	LIU H , LIU Z Y , JIA W Q , et al. Remaining useful life prediction using a novel feature-attention-based end-to-end approach[J]. IEEE Trans.on Industrial Informatics, 2021, 17 (2): 1197- 1207. doi: 10.1109/TII.2020.2983760
12	GUO L , LI N P , JIA F , et al. A recurrent neural network based health indicator for remaining useful life prediction of bearings[J]. Neurocomputing, 2017, 240, 98- 109. doi: 10.1016/j.neucom.2017.02.045
13	TANG G, ZHOU Y G, WANG H Q, et al. Prediction of bearing performance degradation with bottleneck feature based on LSTM network[C]//Proc. of the IEEE International Instrumentation and Measurement Technology Conference, 2018.
14	郑小霞, 钱轶群, 王帅. 基于优选小波包与马氏距离的滚动轴承性能退化GRU预测[J]. 振动与冲击, 2020, 39 (17): 39-46, 63. doi: 10.13465/j.cnki.jvs.2020.17.006
	ZHENG X X , QIAN Y Q , WANG S . Prediction of rolling bearing performance degradation GRU based on optimal wavelet packet and mahalanobis distance[J]. Vibration and Shock, 2020, 39 (17): 39-46, 63. doi: 10.13465/j.cnki.jvs.2020.17.006
15	WU Z H , HUANG N E . Ensemble empirical mode decomposition: a noise-assisted data analysis method[J]. Advances in Adaptive Data Analysis, 2009, 1 (1): 1- 41. doi: 10.1142/S1793536909000047
16	AMAROUAYACHE I I E , SAADI M N , GUERSI N , et al. Bearing fault diagnostics using EEMD processing and convolutional neural network methods[J]. The International Journal of Advanced Manufacturing Technology, 2020, 107 (9): 4077- 4095.
17	MIAO Y H , ZHAO M , LIN J , et al. Application of an improved maximum correlated kurtosis deconvolution method for fault diagnosis of rolling element bearings[J]. Mechanical Systems and Signal Processing, 2017, 92, 173- 195. doi: 10.1016/j.ymssp.2017.01.033
18	YAN X A , JIA M P . A novel optimized SVM classification algorithm with multi-domain feature and its application to fault diagnosis of rolling bearing[J]. Neurocomputing, 2018, 313, 47- 64. doi: 10.1016/j.neucom.2018.05.002
19	REN L , CUI J , SUN Y Q , et al. Multi-bearing remaining useful life collaborative prediction: a deep learning approach[J]. Journal of Manufacturing Systems, 2017, 43, 248- 256. doi: 10.1016/j.jmsy.2017.02.013
20	何正友, 蔡玉梅, 钱清泉. 小波熵理论及其在电力系统故障检测中的应用研究[J]. 中国电机工程学报, 2005, 25 (5): 40- 45.
	HE Z Y , CAI Y M , QIAN Q Q . A study of wavelet entropy theory and its application in electric power system fault detection[J]. Chinese Journal of Electrical Engineering, 2005, 25 (5): 40- 45.
21	IMANI M . Difference-based target detection using Mahalanobis distance and spectral angle[J]. International Journal of Remote Sensing, 2019, 40 (3/4): 811- 831.
22	蒋焕文, 孙续. 电子测量[M]. 北京: 中国计量出版社, 1989.
	JIANG H W , SUN X . Electronic measurement[M]. Beijing: China Metrology Press, 1989.
23	蔺瑞管, 王华伟, 车畅畅, 等. 基于LSTM分类器的航空发动机预测性维护模型[J]. 系统工程与电子技术, 2022, 44 (3): 1052- 1059.
	LIN R G , WANG H W , CHE C C , et al. A predictive maintenance model for aircraft engines based on LSTM classifier[J]. Systems Engineering and Electronics, 2022, 44 (3): 1052- 1059.
24	HE X F, CAI D, NIYOGI P. Laplacian score for feature selection[C]//Proc. of the 18th International Conference on Neural Information Processing Systems, 2005: 507-514.
25	霍超颖, 闫华, 冯雪健, 等. HRRP稀疏自编码器深层特征与散射中心特征的关联性研究[J]. 系统工程与电子技术, 2021, 43 (11): 3040- 3053.
	HUO C Y , YAN H , FENG X J , et al. Research on the correlation between deep features and scattering center features of HRRP sparse autoencoder[J]. Systems Engineering and Electronics, 2021, 43 (11): 3040- 3053.
26	孙文珺, 邵思羽, 严如强. 基于稀疏自动编码深度神经网络的感应电动机故障诊断[J]. 机械工程学报, 2016, 52 (9): 65- 71.
	SUN W J , SHAO S Y , YAN R Q . Fault diagnosis of induction motors based on sparse automatic encoding deep neural networks[J]. Journal of Mechanical Engineering, 2016, 52 (9): 65- 71.
27	SHAO H D , CHENG J S , JIANG H K , et al. Enhanced deep gated recurrent unit and complex wavelet packet energy moment entropy for early fault prognosis of bearing[J]. Know-ledge-based Systems, 2020, 188, 105022.
28	NECTOUX P, GOURIVEAU R, MEDJAHER K, et al. PRONOSTIA: an experimental platform for bearings accelerated degradation tests[C]//Proc. of the IEEE International Conference on Prognostics and Health Management, 2012.
29	LOUTAS T H , ROULIAS D , GEORGOULAS G . Remaining useful life estimation in rolling bearings utilizing data-driven probabilistic E-support vectors regression[J]. IEEE Trans.on Reliability, 2013, 62 (4): 821- 832.

预测类别	真实类别
预测类别	正例	负例
正例	TP	FP
负例	FN	TN

指标	结果
准确率	0.987 6
精确率	0.992 9
灵敏度	0.988 6
特异度	0.986 0
F1-Score	0.990 7

超参数	取值范围
初始学习率	[0.01, 0.005, 0.001]
隐层神经元数	[64,128]
最大迭代次数	[50,100,150]

初始学习率	隐层神经元数	最大迭代次数	RMSE
0.005	128	150	0.414 2
0.005	128	100	0.762 1
0.010	64	150	1.249 6
0.001	128	150	1.357 3
0.005	128	50	1.550 4

网络	超参数	取值
SAE	编码/解码函数	饱和线性参数
	隐层神经元数	6
	L2权重正则化	0.01
	稀疏比例	0.001
GRU	优化函数	自适应矩估计
	初始学习率	0.005
	隐层神经元数	128
	GRU层数	1
	最大迭代次数	150