基于LSTM-DBN的航空发动机剩余寿命预测

doi:10.3969/j.issn.1001-506X.2020.07.28

摘要/Abstract

摘要：

针对航空发动机剩余寿命(remaining useful life, RUL)预测中多传感器监测数据维度高、规模大以及时间序列信息考虑不充分等问题,提出一种融合长短时记忆(long short-term memory, LSTM)网络和深度置信网络(deep belief network, DBN)的RUL预测方法。首先,利用LSTM分别对单一传感器进行时间序列预测。其次,将预测结果整合输入到DBN进行健康指标提取。再次,结合健康指标预测曲线和失效阈值得到RUL预测结果。最后,利用商用模块化航空推进系统仿真数据集开展实验,并与已有方法进行对比分析,验证了该方法的可行性和有效性。

关键词: 长短时记忆网络, 深度置信网络, 健康指标, 剩余寿命预测

Abstract:

In order to solve the problems of high dimension and large scale of multi-sensors monitoring data and insufficient consideration of time series information in the remaining useful life (RUL) prediction of the aircraft engine, a RUL prediction method based on the long short-term memory (LSTM) network and deep belief network (DBN) is proposed. Firstly, the time series of a single sensor is predicted by using the LSTM network. Secondly, the prediction results are integrated into the DBN to extract the health indicator. Thirdly, the RUL prediction results are obtained by combining the health indicator prediction curve and the failure threshold. Finally, to validate the feasibility and the effectiveness of the proposed method, an experiment is carried out on the commercial modular aero-propulsion system simulation data set and the prediction results are compared with the existing methods.

Key words: long short-term memory (LSTM) network, deep belief network (DBN), health indicator, remaining useful life (RUL) prediction

中图分类号:

TP183
V239

李京峰, 陈云翔, 项华春, 蔡忠义. 基于LSTM-DBN的航空发动机剩余寿命预测[J]. 系统工程与电子技术, 2020, 42(7): 1637-1644.

Jingfeng LI, Yunxiang CHEN, Huachun XIANG, Zhongyi CAI. Remaining useful life prediction for aircraft engine based on LSTM-DBN[J]. Systems Engineering and Electronics, 2020, 42(7): 1637-1644.

图/表 14

图1

图2

图3

图4

表1

图5

图6

图7

图8

图9

表2

表3

表4

表5

参考文献 23

1	赵广社, 吴思思, 荣海军. 多源统计数据驱动的航空发动机剩余寿命预测方法[J]. 西安交通大学学报, 2017, 51 (11): 150- 155.
	ZHAO G S , WU S S , RONG H J . A multi-source statistics data-driven method for remaining useful life prediction of aircraft engine[J]. Journal of Xi'an Jiaotong University, 2017, 51 (11): 150- 155.
2	张彬.数据驱动的机械设备性能退化建模与剩余寿命预测研究[D].北京:北京科技大学, 2016.
	ZHANG B. Research on data-driven performance degradation modelling remaining useful life prediction for mechanical equipments[D]. Beijing: University of Science and Technology Beijing, 2016.
3	任子强, 司小胜, 胡昌华, 等. 融合多传感器数据的发动机剩余寿命预测方法[J]. 航空学报, 2019, 40 (12): 134- 145.
	REN Z Q , SI X S , HU C H , et al. Remaining useful life prediction method for engine combining multi-sensors data[J]. Acta Aeronautica et Astronautica Sinica, 2019, 40 (12): 134- 145.
4	罗斌.基于结构疲劳寿命预测的机队维修决策方法研究[D].哈尔滨:哈尔滨工业大学, 2018.
	LUO B. Research on fleet maintenance decision-making based on structure fatigue life prediction[D]. Harbin: Harbin Institute of Technology, 2018.
5	彭开香,皮彦婷,焦瑞华,等.航空发动机的健康指标构建与剩余寿命预测[EB/OL].[2020-01-02].http://kns.cnki.net/kcms/detail/44.1240.TP.20190923.2124.014.html.
	PENG K X, PI Y T, JIAO R H, et al. Health indicator construction and remaining useful life prediction for aircraft engine[EB/OL].[2020-01-02].http://kns.cnki.net/kcms/detail/44.1240.TP.20190923.2124.014.html.
6	EL-THALJI I , JANTUNEN E . A summary of fault modelling and predictive health monitoring of rolling element bearings[J]. Mechanical Systems and Signal Processing, 2015, 60-61, 252- 272.
7	HU C H , HONG P , WANG Z Q , et al. A new remaining useful life estimation method for equipment subjected to intervention of imperfect maintenance activities[J]. Chinese Journal of Aeronautics, 2018, 31 (3): 514- 528.
8	ZHANG H W , CHEN M Y , ZHOU D H . Remaining useful life prediction for a nonlinear multi-degradation system with public noise[J]. Journal of Systems Engineering and Electronics, 2018, 29 (2): 429- 435.
9	马小骏, 任淑红, 左洪福, 等. 基于LS-SVM算法和性能可靠性的航空发动机在翼寿命预测方法[J]. 交通运输工程学报, 2015, 15 (3): 92- 100.
	MA X J , REN S H , ZUO H F , et al. Prediction method of aero-engine life on wing based on LS-SVM algorithm and performance reliability[J]. Journal of Traffic and Transportation Engineering, 2015, 15 (3): 92- 100.
10	TANG D Y , CAO J R , YU J S . Remaining useful life prediction for engineering systems under dynamic operational conditions: a semi-Markov decision process-based approach[J]. Chinese Journal of Aeronautics, 2019, 32 (3): 627- 638.
11	XIE P . Analysis of fault of insulation aging of oiled paper of a large-scale power transformer and the prediction of its service life[J]. IEEE Trans.on Electrical and Electronic Engineering, 2019, 14 (8): 1139- 1144.
12	KHAN S , YAIRI T . A review on the application of deep learning in system health management[J]. Mechanical Systems and Signal Processing, 2018, 107, 241- 265.
13	CHE C C , WANG H W , FU Q , et al. Combining multiple deep learning algorithms for prognostic and health management of aircraft[J]. Aerospace Science and Technology, 2019, 94, 105423.
14	ZHENG S, RISTOVSKI K, FARAHAT A, et al. Long short-term memory network for remaining useful life estimation[C]//Proc.of the IEEE International Conference on Prognostics and Health Management, 2017: 88-95.
15	ZHANG J J , WANG P , YAN R Q , et al. Long short-term memory for machine remaining life prediction[J]. Journal of Manufacturing Systems, 2018, 48 (SI): 78- 86.
16	HOCHREITER S , SCHMIDHUBER J . Long short-term memory[J]. Neural Computation, 1997, 9 (8): 1735- 1780.
17	HINTON G , OSINDERO S , TEH Y . A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18 (7): 1527- 1554.
18	HINTON G . A practical guide to training restricted boltzmann machines[J]. Momentum, 2010, 9 (1): 926- 947.
19	HINTON G . Training products of experts by minimizing contrastive divergence[J]. Neural Computation, 2002, 14 (8): 1771- 1800.
20	SRIVASTAVA N , HINTON G , KRIZHEVSKY A , et al. Dropout: a simple way to prevent neural networks from overfitting[J]. Journal of Machine Learning Research, 2014, 15, 1929- 1958.
21	BABU G S, ZHAO P L, LI X L. Deep convolutional neural network based regression approach for estimation of remaining useful life[C]//Proc.of the International Conference on Database Systems for Advanced Applications, 2016: 214-228.
22	SAXENA A, KAI G, SIMON D, et al. Damage propagation modeling for aircraft engine run-to-failure simulation[C]//Proc.of the International Conference on Prognostics and Health Management, 2008.
23	宋亚, 夏唐斌, 郑宇, 等. 基于Autoencoder-BLSTM的涡扇发动机剩余寿命预测[J]. 计算机集成制造系统, 2019, 25 (7): 1611- 1619.
	SONG Y , XIA T B , ZHENG Y , et al. Remaining useful life of turbofan engine based on Autoencoder-BLSTM[J]. Computer Integrated Manufacturing Systems, 2019, 25 (7): 1611- 1619.

参数	取值
参数	LSTM模型	DBN模型
网络结构	-	[14,9,5,1]
批尺寸	80	30
BP训练时期数	150	60
RBM训练时期数	-	50
BP学习率	0.001	0.1
RBM学习率	-	0.1

参数	数值	57号发动机	89号发动机	93号发动机
真实失效周期	-	148	147	146
起始点55	失效周期	115	94	129
起始点55	预测误差	-33	-53	-17
起始点100	失效周期	117	113	131
起始点100	预测误差	-31	-34	-15
起始点145	失效周期	148	146	146
起始点145	预测误差	0	-1	0

设置	取值
设置	Score	RMSE
起始点55	72.32	37.36
起始点100	24.70	27.94
起始点145	0.08	0.58

预测方法	FD001	FD002	FD003	FD004
MLP	17 972	7 802 800	17 409	5 616 600
SVR	1 382	589 900	1 598	371 140
CNN	1 287	13 570	1 596	7 886
深度LSTM	338	4 450	852	5 550
自编码BLSTM	261	-	-	-
LSTM-DBN	172	1 937	226	2 792

预测方法	FD001	FD002	FD003	FD004
MLP	37.56	80.03	37.39	77.37
SVR	20.96	42.00	21.05	45.35
CNN	18.45	30.29	19.82	29.16
深度LSTM	16.14	24.49	16.18	28.17
自编码BLSTM	13.63	-	-	-
LSTM-DBN	12.43	19.89	12.65	20.51