基于时序图推理的设备剩余使用寿命预测

刘雨蒙; 郑旭; 田玲; 王宏安

doi:10.16383/j.aas.c230014

基于时序图推理的设备剩余使用寿命预测

doi: 10.16383/j.aas.c230014

刘雨蒙^{1, 2, 3,},
郑旭^4,,
田玲^4,,
王宏安^{1, 2,}

1.
中国科学院软件研究所人机交互北京市重点实验室北京 100190
2.
中国科学院大学北京 100049
3.
中国科学院软件研究所天基综合信息系统重点实验室北京 100190
4.
电子科技大学计算机科学与工程学院成都 611731

基金项目: 四川省科技计划(重点研发) (2021YFG0018, 2022YFG0038), 国防基础研究计划 (JCKY2020903B002)资助

详细信息

作者简介:
刘雨蒙：中国科学院软件研究所高级工程师. 2017年获得美国南加利福尼亚大学硕士学位. 主要研究方向为数据挖掘, 数据库技术. E-mail: yumeng@iscas.ac.cn

郑旭：电子科技大学计算机科学与工程学院副教授. 2018年获得美国佐治亚州立大学计算机科学系博士学位. 主要研究方向为物联网, 人工智能. 本文通信作者. E-mail: xzheng@uestc.edu.cn

田玲：电子科技大学计算机科学与工程学院教授. 2010年获得电子科技大学博士学位. 主要研究方向为人工智能. E-mail: ruan052@126.com

王宏安：中国科学院软件研究所研究员. 1998年获得中国科学院软件研究所博士学位. 主要研究方向为人机交互, 实时智能. E-mail: hongan@iscas.ac.cn

计量
- 文章访问数: 609
- HTML全文浏览量: 732
- PDF下载量: 335
- 被引次数: 0
出版历程
- 收稿日期: 2023-01-12
- 录用日期: 2023-06-14
- 网络出版日期: 2023-09-18
- 刊出日期: 2024-01-29

Remaining Useful Life Estimation of Facilities Based on Reasoning Over Temporal Graphs

LIU Yu-Meng^{1, 2, 3
,},
ZHENG Xu^4
,,
TIAN Ling^4
,,
WANG Hong-An^{1, 2
,}

1.
Beijing Key Laboratory of Human-computer Interaction, Institute of Software, Chinese Academy of Sciences, Beijing 100190
2.
University of Chinese Academy of Sciences, Beijing 100049
3.
Science and Technology on Integrated Information System Laboratory, Institute of Software, Chinese Academy of Sciences, Beijing 100190
4.
School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731

Funds: Supported by Sichuan Science and Technology Program (Key Research and Development Program) (2021YFG0018, 2022YFG0038) and National Defense Basic Scientific Research Program of China (JCKY2020903B002)

More Information

Author Bio:
LIU Yu-Meng　Senior engineer at the Institute of Software, Chinese Academy of Sciences. He received his master degree from University of Southern California in 2017. His research interest covers data mining and database technology

ZHENG Xu　Associate professor at the School of Computer Science and Engineering, University of Electronic Science and Technology of China. He received his Ph.D. degree in computer science from Georgia State University in 2018. His research interest covers internet of things and artificial intelligence. Corresponding author of this paper

TIAN Ling　Professor at the School of Computer Science and Engineering, University of Electronic Science and Technology of China. She received her Ph.D. degree from University of Electronic Science and Technology of China in 2010. Her main research interest is artificial intelligence

WANG Hong-An　Researcher at the Institute of Software, Chinese Academy of Sciences. He received his Ph.D. degree from the Institute of Software, Chinese Academy of Sciences in 1998. His research interest covers human-computer interaction and real-time intelligence

摘要

摘要: 剩余使用寿命(Remaining useful life, RUL)预测是大型设备故障预测与健康管理(Prognostics and health management, PHM)的重要环节, 对于降低设备维修成本和避免灾难性故障具有重要意义. 针对RUL预测, 首次提出一种基于多变量分析的时序图推理模型(Multivariate similarity temporal knowledge graph, MSTKG), 通过捕捉设备各部件的运行状态耦合关系及其变化趋势, 挖掘其中蕴含的设备性能退化信息, 为寿命预测提供有效依据. 首先, 设计时序图结构, 形式化表达各部件不同工作周期的关联关系. 其次, 提出联合图卷积神经网络(Convolutional neural network, CNN)和门控循环单元 (Gated recurrent unit, GRU)的深度推理网络, 建模并学习设备各部件工作状态的时空演化过程, 并结合回归分析, 得到剩余使用寿命预测结果. 最后, 与现有预测方法相比, 所提方法能够显式建模并利用设备部件耦合关系的变化信息, 仿真实验结果验证了该方法的优越性.
- 剩余使用寿命 /
- 时序图推理 /
- 图神经网络 /
- 深度推理网络
Abstract: Remaining useful life (RUL) estimation is an important component of the prognostics and health management (PHM) of large-scale equipment, which is of great significance for equipment maintenance and avoiding catastrophic failures. In this paper, a multivariate similarity temporal knowledge graph model (MSTKG) is proposed for remaining useful life evaluation. The model can capture the dynamic information such as time-evolving changes in the coupling relationship and stability of the various components of the equipment, and mine the equipment performance degradation information accordingly. Firstly, a temporal and sequential graph structure is designed to capture the relationships among multiple sensors in adjacent time period, and to formally represent the evolving correlations among different components in continuous working cycle. Secondly, we propose a deep inference network which integrates relation-aware convolutional neural network (CNN) and gated recurrent unit (GRU). The inference network explicitly learns the spatial and temporal evolution of the state of each component of the equipment. The results of remaining useful life estimation are obtained by integrating regression analysis. Finally, compared with existing estimation methods, the proposed method is able to explicitly model and utilize the changing information of equipment component coupling relationships. Extensive evaluation results on benchmark show that the proposed model outperforms previous solutions on RUL estimation.
- Remaining useful life (RUL) /
- temporal graph reasoning /
- graph neural network /
- deep inference network

HTML全文

图 1 MSTKG模型结构图

Fig. 1 Structure of MSTKG model

下载: 全尺寸图片幻灯片

图 2 DTW示意图

Fig. 2 Illustration of DTW

下载: 全尺寸图片幻灯片

图 3 FD001 1号发动机传感器监测数据

Fig. 3 Sensor monitoring data for engine unit 1 of FD001

下载: 全尺寸图片幻灯片

图 4 FD001 ~ FD004 RUL预测结果

Fig. 4 Prediction results of RUL on FD001 ~ FD004

下载: 全尺寸图片幻灯片

图 5 FD003中39号和47号发动机单元预测结果分析

Fig. 5 Prediction results analysis of unit 39 and unit 47 engines on FD003

下载: 全尺寸图片幻灯片

图 6 FD004 229号发动机单元关联依赖分析

Fig. 6 Association dependence analysis of unit 229 engine on FD004

下载: 全尺寸图片幻灯片

表 1 CMAPSS传感器数据描述

Table 1 The description of CMAPSS sensor data

传感器编号	简称	描述
1	T2	风扇进口温度
2	T24	低压压气机出口温度
3	T30	高压压气机出口温度
4	T50	低压涡轮出口温度
5	P2	风扇出口压力
6	P15	涵道压力
7	P30	高压压气机出口压力
8	Nf	物理风扇转速
9	Nc	核心转速
10	Epr	发动机增压比
11	Ps30	高压压气机出口静压
12	phi	燃料流量比
13	NRf	校正后风扇转速
14	NRc	校正后核心转速
15	BPR	涵道比
16	farB	燃烧空气比
17	htBleed	排气焓
18	Nf_dmd	要求风扇转速
19	PCNfR_dmd	要求风扇校正转速
20	W31	高压涡轮冷却液流速
21	W32	低压涡轮冷却液流失

下载: 导出CSV

表 2 CMAPSS数据集中4个子集的细节信息

Table 2 Detailed information of four subsets of CMAPSS dataset

	运行状态数	故障模式数	传感器个数	训练单元个数	测试单元个数
FD001	1	1	21	100	100
FD002	6	1	21	260	259
FD003	1	2	21	100	100
FD004	6	2	21	249	248

下载: 导出CSV

表 3 CMAPSS数据集实验性能对比

Table 3 Comparison of experimental performance on the CMAPSS dataset

对比方法	FD001		FD002		FD003		FD004		平均
对比方法	RMSE	Score	RMSE	Score	RMSE	Score	RMSE	Score	RMSE	Score
CNN (2016)^[18]	18.45	—	30.29	—	19.82	—	29.16	—	26.74	—
LSTM-FNN^[22]	16.14	338.00	24.49	4450.00	16.18	852.00	28.17	5550.00	23.42	3745.33
CNN-FNN^[19]	12.61	274.00	22.36	10412.00	12.64	284.00	22.43	12466.00	19.63	8266.02
Autoencoder^[34]	14.74	273.00	22.07	3099.00	17.48	574.00	23.49	3202.00	20.88	2378.27
RBM-LSTM-FNN^[30]	12.56	231.00	22.73	3366.00	12.10	251.00	22.66	2840.00	19.76	2297.47
DCNN-FNN^[33]	12.61	—	28.51	—	12.62	—	30.73	—	24.79	—
MODBNE^[35]	15.04	334.00	25.05	5585.00	12.51	422.00	28.66	6558.00	23.13	4453.32
DBN^[36]	15.04	334.00	25.05	5585.00	12.51	421.00	28.66	6557.00	23.13	4452.83
RULCLIPPER^[37]	13.27	216.00	22.89	2796.00	17.48	574.00	23.49	3202.00	20.97	2259.21
Autoencoder^[21]	13.58	228.00	19.59	2650.00	19.16	1727.00	22.15	2901.00	19.58	2264.92
GCU-Transformer^[29]	11.27	—	22.81	—	11.42	—	24.86	—	20.29	—
本文提出的方法	16.69	497.49	18.70	1605.91	17.48	651.70	20.86	2384.60	19.00	1587.31

下载: 导出CSV

表 4 阶段性RUL预测均方根误差

Table 4 Phased RUL prediction RMSE

阶段	FD001	FD002	FD003	FD004	平均
总体	16.69	18.71	17.48	20.86	18.44
前30%	19.57	23.11	19.85	23.86	21.60
后70%	15.30	16.48	16.37	19.44	16.90

下载: 导出CSV

表 5 FD002数据子集上对不同结构的消融研究

Table 5 Experimental ablation study on different structures on FD002

模型结构	RMSE	性能降低
Ours	20.86	—
Ours w/o relation evolution	23.40	−2.54
Ours w/o relation	23.04	−2.18
Ours w/o origin input	21.85	−0.99

下载: 导出CSV

表 6 不同参数设置的模型预测性能

Table 6 Model prediction performance with different parameter settings

时间窗口长度	时间窗口跨度	时间窗口个数	RMSE
5	1	1	24.34
5	1	5	21.37
5	1	7	20.85
5	1	10	22.06
5	3	7	22.81

下载: 导出CSV

参考文献(37)

[1]	Zhong M Y, Ding S X, Zhou D H, He X. An H_i/H_∞ optimization approach to event-triggered fault detection for linear discrete time systems. IEEE Transactions on Automatic Control, 2020, 65(10): 4464-4471 doi: 10.1109/TAC.2020.3006811
[2]	Sun Y J, Wang J, Wang X H. Fault diagnosis of mechanical equipment in high energy consumption industries in China: A review. Mechanical Systems and Signal Processing, 2023, 186: Article No. 109833. doi: 10.1016/j.ymssp.2022.109833
[3]	裴洪, 胡昌华, 司小胜, 张正新, 杜党波. 不完美维护下基于剩余寿命预测信息的设备维护决策模型. 自动化学报, 2018, 44(4): 719-729 Pei Hong, Hu Chang-Hua, Si Xiao-Sheng, Zhang Zheng-Xin, Du Dang-Bo. Remaining life prediction information-based maintenance decision model for equipment under imperfect maintenance. Acta Automatica Sinica, 2018, 44(4): 719-729
[4]	Camci F, Chinnam R B. Health-state estimation and prognostics in machining processes. IEEE Transactions on Automation Science and Engineering, 2010, 7(3): 581-597 doi: 10.1109/TASE.2009.2038170
[5]	Liao L, Kottig F. Review of hybrid prognostics approaches for remaining useful life prediction of engineered systems, and an application to battery life prediction. IEEE Transactions on Reliability, 2014, 63(1): 191-207 doi: 10.1109/TR.2014.2299152
[6]	陆宁云, 陈闯, 姜斌, 邢尹. 复杂系统维护策略最新研究进展: 从视情维护到预测性维护. 自动化学报, 2021, 47(1): 1-17 Lu Ning-Yun, Chen Chuang, Jiang Bin, Xing Yin. Latest progress on maintenance strategy of complex system: From condition-based maintenance to predictive maintenance. Acta Automatica Sinica, 2021, 47(1): 1-17
[7]	Souza A D R D, Efimov D, Raissi T, Ping X B. Robust output feedback model predictive control for constrained linear systems via interval observers. Automatica, 2022, 135: Article No.109951 doi: 10.1016/j.automatica.2021.109951
[8]	Wang B, Lei Y G, Li N P, Li N B. A hybrid prognostics approach for estimating remaining useful life of rolling element bearings. IEEE Transactions on Reliability, 2020, 69(1): 401-412 doi: 10.1109/TR.2018.2882682
[9]	袁烨, 张永, 丁汉. 工业人工智能的关键技术及其在预测性维护中的应用现状. 自动化学报, 2020, 46(10): 2013-2030 Yuan Ye, Zhang Yong, Ding Han. Research on key technology of industrial artificial intelligence and its application in predictive maintenance. Acta Automatica Sinica, 2020, 46(10): 2013-2030
[10]	董青, 郑建飞, 胡昌华, 李冰, 牟含笑. 基于两阶段自适应Wiener过程的剩余寿命预测方法. 自动化学报, 2022, 48(2): 539-553 Dong Qing, Zheng Jian-Fei, Hu Chang-Hua, Li Bing, Mu Han-Xiao. Remaining useful life prognostic method based on two-stage adaptive wiener process. Acta Automatica Sinica, 2022, 48(2): 539-553
[11]	Wang W L, Gong C H, Wang D. Optimizing sensor activation in a Language domain for fault diagnosis. IEEE Transactions on Automatic Control, 2019, 64(2) 743-750 doi: 10.1109/TAC.2018.2844409
[12]	Li Q, Wang Z D, Hu J, Sheng W G. Distributed state and fault estimation over sensor networks with probabilistic quantizations: The dynamic event-triggered case. Automatica, 2021, 131: Article No.109784 doi: 10.1016/j.automatica.2021.109784
[13]	李天梅, 司小胜, 刘翔, 裴洪. 大数据下数模联动的随机退化设备剩余寿命预测技术. 自动化学报, 2022, 48(9): 2119-2141 Li Tian-Mei, Si Xiao-Sheng, Liu Xiang, Pei Hong. Data-model interactive remaining useful life prediction technologies for stochastic degrading devices with big data. Acta Automatica Sinica, 2022, 48(9): 2119-2141
[14]	李杰, 李响, 许元铭, 杨绍杰, 孙可意. 工业人工智能及应用研究现状及展望. 自动化学报, 2020, 46(10): 2031-2044 Li Jie, Li Xiang, Xu Yuan-Ming, Yang Shao-Jie, Sun Ke-Yi. Recent advances and prospects in industrial AI and applications. Acta Automatica Sinica, 2020, 46(10): 2031-2044
[15]	韩中, 程林, 熊金泉, 刘满军. 大数据结构化与数据驱动的复杂系统维修决策. 自动化学报, 2020, 46(2): 385-396 Han Zhong, Cheng Lin, Xiong Jin-Quan, Liu Man-Jun. Complex system maintenance decisions based on big data structuration and data-driven. Acta Automatica Sinica, 2020, 46(2): 385-396
[16]	喻勇, 司小胜, 胡昌华, 崔忠马, 李洪鹏. 数据驱动的可靠性评估与寿命预测研究进展: 基于协变量的方法. 自动化学报, 2018, 44(2): 216-227 Yu Yong, Si Xiao-Sheng, Hu Chang-Hua, Cui Zhong-Ma, Li Hong-Peng. Data driven reliability assessment and life-time prognostics: A review on covariate models. Acta Automatica Sinica, 2018, 44(2): 216-227
[17]	Yang B Y, Liu R N, Zio E. Remaining useful life prediction based on a double-convolutional neural network architecture. IEEE Transactions on Industrial Electronics, 2019, 66(12): 9521-9530 doi: 10.1109/TIE.2019.2924605
[18]	Babu G S, Zhao P L, Li X L. Deep convolutional neural network based regression approach for estimation of remaining useful life. In: Proceedings of the International Conference on Database Systems for Advanced Applications. Cham, Switzerland: Springer, 2016. 214–228
[19]	Li X, Ding Q, Sun J Q. Remaining useful life estimation in prognostics using deep convolution neural networks. Reliability Engineering System Safety, 2018, 172: 1-11. doi: 10.1016/j.ress.2017.11.021
[20]	Malhi A, Yan R Q, Gao R X. Prognosis of defect propagation based on recurrent neural networks. IEEE Transactions on Instrumentation and Measurement, 2011, 60(3): 703-711 doi: 10.1109/TIM.2010.2078296
[21]	Yu W N, Kim I Y, Mechefske C. An improved similarity-based prognostic algorithm for RUL estimation using an RNN autoencoder scheme. Reliability Engineering System Safety, 2020, 199: Article No. 106926. doi: 10.1016/j.ress.2020.106926
[22]	Zheng S, Ristovski K, Farahat A, Gupta C. Long short-term memory network for remaining useful life estimation. In: Proceedings of the IEEE International Conference on Prognostics and Health Management (ICPHM). Dallas, USA: IEEE, 2017. 88−95
[23]	Jin R B, Chen Z H, Wu K Y, Wu M, Li X L, Yan R Q. Bi-LSTM based two-stream network for machine remaining useful life prediction. IEEE Transactions on Instrumentation and Measurement, 2022, 71: 1-10.
[24]	Liu C L, Zhang Y, Sun J R, Cui Z H, Wang K. Stacked bidirectional LSTM RNN to evaluate the remaining useful life of supercapacitor, International Journal of Energy Research, 2022, 46(3): 3034–3043. doi: 10.1002/er.7360
[25]	康守强, 周月, 王玉静, 谢金宝, MIKULOVICH Vladimir Ivanovich. 基于改进SAE和双向LSTM的滚动轴承RUL预测方法. 自动化学报, 2022, 48(9): 2327-2336 Kang Shou-Qiang, Zhou Yue, Wang Yu-Jing, Xie Jin-Bao, MIKULOVICH Vladimir Ivanovich. RUL prediction method of a rolling bearing based on improved SAE and BI-LSTM. Acta Automatica Sinica, 2022, 48(9): 2327-2336
[26]	Chen J G, Chen D J, Liu G P.. Using temporal convolution network for remaining useful lifetime prediction. Engineering Reports, 2021, 3(3): Article No. e12305. doi: 10.1002/eng2.12305
[27]	Chang Y H, Chen J L, Liu Y L, Xu E Y, He S L. Temporal convolution-based sorting feature repeat-explore network combining with multi-band information for remaining useful life estimation of equipment. Knowledge-based Systems, 2022, 249: Article No. 108958. doi: 10.1016/j.knosys.2022.108958
[28]	Liu C L, Li D Z, Wang L C, Li L W, Wang K. Strong robustness and high accuracy in predicting remaining useful life of supercapacitors. APL Materials, 2022, 10(6): Article No. 061106.
[29]	Mo Y, Wu Q H, Li X, Huang B Q. Remaining useful life estimation via transformer encoder enhanced by a gated convolutional unit. Journal of Intelligent Manufacturing, 2021, 32(7): 1997-2006 doi: 10.1007/s10845-021-01750-x
[30]	Chen D, Hong W, Zhou X. Transformer network for remaining useful life prediction of lithium-ion batteries. IEEE Access, 2022, 10: 19621-19628 doi: 10.1109/ACCESS.2022.3151975
[31]	Zhang Z Z, Song W, Li Q Q. Dual-aspect self-attention based on transformer for remaining useful life prediction.IEEE Transactions on Instrumentation and Measurement, 2022, 71: 1-11
[32]	Ahmadzadeh F, Lundberg J. Remaining useful life estimation: Review. International Journal of System Assurance Engineering and Management, 2014, 5(4): 461-474
[33]	Xu X, Wu Q, Li X, Huang B. Dilated convolution neural network for remaining useful life prediction. Journal of Computing and Information Science in Engineering, 2020, 2: 021004
[34]	Yu W N, Kim I Y, Mechefske C. Remaining useful life estimation using a bidirectional recurrent neural network based autoencoder scheme. Mechanical Systems and Signal Processing, 2019, 129: 764–780 doi: 10.1016/j.ymssp.2019.05.005
[35]	Zhang C, Lim P, Qin A K, Tan K C. Multiobjective deep belief networks ensemble for remaining useful life estimation in prognostics. IEEE Transactions on Neural Networks and Learning Systems, 2017, 28(10): 2306–2318. doi: 10.1109/TNNLS.2016.2582798
[36]	Liao L X, Jin W J, Pavel R. Enhanced restricted boltzmann machine with prognosability regularization for prognostics and health assessment. IEEE Transactions on Industrial Electronics, 2016, 63(11): 7076–7083. doi: 10.1109/TIE.2016.2586442
[37]	Cai H S, Feng J S, Li W Z, Hsu Y M, Lee J. Similarity-based particle filter for remaining useful life prediction with enhanced performance. Applied Soft Computing, 2020, 94: Article No. 106474. doi: 10.1016/j.asoc.2020.106474