基于RCNN-ABiLSTM的机械设备剩余寿命预测方法

doi:10.12305/j.issn.1001-506X.2023.03.35

摘要/Abstract

摘要：

针对机械设备的关键退化信息易淹没在非线性、多维度、长时间、大规模监测数据中的问题, 提出了一种基于残差卷积神经网络和注意力双向长短时记忆网络融合(residual convolutional neural network-attentional bidirectional long short-term memory network, RCNN-ABiLSTM)的机械设备剩余寿命预测方法。首先通过训练RCNN提取监测数据的深度空间特征; 然后通过引入注意力机制, 优化双向长短时记忆网络提取时间相关特征的权重参数, 加强关键退化信息对剩余寿命预测的表达; 最后通过航空发动机数据集验证了方法的有效性。分析结果表明, 对于运行条件复杂和故障模式多变的多维监测数据, 所提方法能够准确寻找退化时间点, 有效提高长时间运行设备的剩余寿命预测准确度。

关键词: 残差卷积神经网络, 注意力机制, 融合模型, 剩余寿命预测, 航空发动机

Abstract:

Aiming at the problem that the key degradation information of mechanical equipment is easy to be submerged in nonlinear, multi-dimensional, long-term and large-scale monitoring data, a method for predicting the remaining useful life of mechanical equipment based on residual convolutional neural network-attentional bidirectional long short-term memory network(RCNN-ABiLSTM) is proposed. Firstly, the RCNN is trained for deep spatial feature extraction of the monitoring data.Then, by introducing the attention mechanism, the weight parameters of the time-related features extracted by BiLSTM are optimized. And the expression of the key degradation information on the remaining life prediction is strengthened. Finally, the effectiveness of the proposed method is verified by the aircraft engine. The analysis results show that the proposed method can accurately find the degradation time point for multi-dimensional monitoring data with complex operating conditions and variable failure modes. The remaining useful life prediction accuracy of long-running equipment is effectively improved.

Key words: residual convolutional neural network (RCNN), attention mechanism, fusion model, remaining useful life (RUL) prediction, aircraft engine

中图分类号:

TP183

闫啸家, 梁伟阁, 张钢, 佘博, 田福庆. 基于RCNN-ABiLSTM的机械设备剩余寿命预测方法[J]. 系统工程与电子技术, 2023, 45(3): 931-940.

Xiaojia YAN, Weige LIANG, Gang ZHANG, Bo SHE, Fuqing TIAN. Prediction method for mechanical equipment based on RCNN-ABiLSTM[J]. Systems Engineering and Electronics, 2023, 45(3): 931-940.

图/表 15

图1

图2

图3

图4

图5

图6

表1

图7

图8

表2

图9

图10

图11

图12

表3

参考文献 30

1	SAIDI L , BENBOUZID M . Prognostics and health management of renewable energy systems: state of the art review, challenges, and trends[J]. Electronics, 2021, 10 (22): 27- 32.
2	CHENG C , MA G J , ZHANG Y , et al. A deep learning-based remaining useful life prediction approach for bearings[J]. IEEE/ASME Trans.on Mechatronics, 2020, 25 (3): 1243- 1254. doi: 10.1109/TMECH.2020.2971503
3	GUO R X , WANG Y , ZHANG H C , et al. Remaining useful life prediction for rolling bearings using EMD-RISI-LSTM[J]. IEEE Trans.on Instrumentation and Measurement, 2021, 70, 3509812.
4	YU W , KIM I Y , MECHEFSKE C . Remaining useful life estimation using a bidirectional recurrent neural network based autoencoder scheme[J]. Mechanical Systems and Signal Processing, 2019, 129 (15): 764- 780.
5	CUBILLO A , VERMEULEN J , RODRIGUEZ D L P M , et al. Physics-based integrated vehicle health management system for predicting the remaining useful life of an aircraft planetary gear transmission[J]. International Journal of Structural Integrity, 2017, 8 (4): 17- 35.
6	FAN Z L , LIU G B , SI X S , et al. Degradation data-driven approach for remaining useful life estimation[J]. Systems Engineering and Electronics, 2013, 1 (10): 173- 182.
7	LI Y X , HUANG X Z , DING P F , et al. Wiener-based remaining useful life prediction of rolling bearings using improved Kalman filtering and adaptive modification[J]. Measurement, 2021, 18 (2): 10- 16.
8	黄亮, 刘君强, 贡英杰. 基于Wiener过程的发动机多阶段剩余寿命预测[J]. 北京航空航天大学学报, 2018, 44 (5): 1081- 1087.
	HUANG L , LIU J Q , GONG Y J . Multi-stage remaining life prediction of engine based on Wiener process[J]. Journal of Beijing University of Aeronautics and Astronautics, 2018, 44 (5): 1081- 1087.
9	袁庆洋, 叶建华, 李晓钢. BLDC电机温度退化多段Wiener过程建模[J]. 北京航空航天大学学报, 2018, 44 (7): 1514- 1519.
	YUAN Q Y , YE J H , LI X G . Multi-stage Wiener process modeling for temperature degradation of BLDC motors[J]. Journal of Beijing University of Aeronautics and Astronautics, 2018, 44 (7): 1514- 1519.
10	冯磊, 王宏力, 司小胜, 等. 基于半随机滤波-期望最大化算法的剩余寿命在线预测[J]. 航空学报, 2015, 36 (2): 555- 563.
	FENG L , WANG H L , SI X S , et al. Online prediction of remaining life based on semi-random filtering-expectation-maximization algorithm[J]. Acta Aeronautica et Astronautica Sinica, 2015, 36 (2): 555- 563.
11	KUNDU P , DARPE A K , KULKARNI M S . Weibull accelerated failure time regression model for remaining useful life prediction of bearing working under multiple operating conditions[J]. Mechanical Systems and Signal Processing, 2019, 134 (1): 106302.
12	ZHANG Y B , JIA Y X , FENG T L , et al. Remaining useful life prediction model of planetary carrier in helicopter main gear-box based on Gamma degradation process[J]. Journal of Vibration and Shock, 2012, 31 (14): 47- 51.
13	王泽洲, 陈云翔, 蔡忠义, 等. 基于比例关系加速退化建模的设备剩余寿命在线预测[J]. 系统工程与电子技术, 2021, 43 (2): 584- 592.
	WANG Z Z , CHEN Y X , CAI Z Y , et al. Equipment remaining useful lifetime online prediction based on accelerated degradation modeling with the proportion relationship[J]. Systems Engineering and Electronics, 2021, 43 (2): 584- 592.
14	WU M H , WANG X M , LIU X Z . On condition maintenance model for complex electromechanical equipments based on remaining useful life and Wiener process[J]. Journal of Physics Conference Series, 2020, 16 (78): 12- 14.
15	BABU G S, ZHAO P, 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.
16	HOCHREITER S , SCHMIDHUBER J , et al. Long short-term memory[J]. Neural Computation, 1997, 9 (8): 1735- 1780.
17	ZHENG S A, 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.
18	王鑫, 吴际, 刘超, 等. 基于LSTM循环神经网络的故障时间序列预测[J]. 北京航空航天大学学报, 2018, 44 (4): 772- 784.
	WANG X , WU J , LIU C , et al. Fault time series prediction based on LSTM recurrent neural network[J]. Journal of Beijing University of Aeronautics and Astronautics, 2018, 44 (4): 772- 784.
19	XIANG S , QIN Y , LUO J , et al. Multicellular LSTM-based deep learning model for aero-engine remaining useful life prediction[J]. Reliability Engineering & System Safety, 2021, 216, 107927.
20	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. Prediction of remaining life of turbofan engine based on Autoencoder-BLSTM[J]. Computer Integrated Manufacturing System, 2019, 25 (7): 1611- 1619.
21	孙世岩, 张钢, 梁伟阁, 等. 基于时间序列数据扩增和BLSTM的滚动轴承剩余寿命预测方法[J]. 系统工程与电子技术, 2022, 44 (3): 1060- 1068.
	SUN S Y , ZHANG G , LIANG W G , et al. Remaining useful life prediction method of rolling bearing based on time series data augmentation and BLSTM[J]. Systems Engineering and Electronics, 2022, 44 (3): 1060- 1068.
22	CHEN J G , CHEN D J , LIU G P . Using temporal convolution network for remaining useful lifetime prediction[J]. Engineering Reports, 2020, 3 (3): 12305.
23	KUO P H , HUANG C J . A high precision artificial neural networks model for short-term energy load forecasting[J]. Energies, 2018, 11 (1): 213- 226.
24	YAO D C , LI B Y , LIU H C , et al. Remaining useful life prediction of roller bearings based on improved 1D-CNN and simple recurrent unit[J]. Measurement, 2021, 175 (4): 10- 16.
25	李永刚, 王朝晖, 万晓依, 等. 基于深度残差双单向DLSTM的时空一致视频事件识别[J]. 计算机学报, 2018, 41 (12): 2852- 2866.
	LI Y G , WANG C H , WAN X Y , et al. Spatiotemporally consistent video event recognition based on deep residual dual unidirectional DLSTM[J]. Chinese Journal of Computers, 2018, 41 (12): 2852- 2866.
26	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proc. of the IEEE Conference on Computer vision, 2016: 770-778.
27	MA X L , TAO Z M , WANG Y H , et al. Long short-term memory neural network for traffic speed prediction using remote microwave sensor data[J]. Transportation Research Part C: Emerging Technologies, 2015, 54, 187- 197.
28	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.
29	HEIMES F O. Recurrent neural networks for remaining useful life estimation[C]//Proc. of the International Conference on Prognostics and Health Management, 2008.
30	KONG Z M , CUI Y D , XIA Z , et al. Convolution and long short-term memory hybrid deep neural networks for remaining useful life prognostics[J]. Applied Sciences, 2019, 9 (19): 41- 56.

发动机	类型	发动机台数	数据总量	工作状态	故障模式
FD001	训练集	100	20 631	1	1
FD001	测试集	100	13 096	1	1
FD002	训练集	260	53 759	6	1
FD002	测试集	259	33 991	6	1
FD003	训练集	100	24 720	1	2
FD003	测试集	100	16 596	1	2
FD004	训练集	249	61 249	6	2
FD004	测试集	248	41 214	6	2

参数	取值	参数	取值
FD001到FD004的S	30/20/30/15	边缘填充	same
RCNN卷积核大小	2×1	优化器	Adam
RCNN的激活函数	tanh	学习率	0.002
RCNN层数	5	批尺寸	128
LSTM神经元数目	64	迭代次数	80
LSTM层数	1	Dropout率	0.25

方法	FD001		FD002		FD003		FD004
方法	RMSE	Score	RMSE	Score	RMSE	Score	RMSE	Score
SVR^[15]	20.96	1 382	35.96	18 900	21.64	2 956	36.54	10 023
CNN^[15]	18.45	1 290	30.29	13 600	19.82	1 600	29.16	7 890
LSTM^[17]	16.14	338	24.49	4 450	16.18	852	28.17	5 550
CNN-LSTM^[30]	16.13	303	20.46	3 440	17.12	1 420	23.26	4 630
Autoencoder-BiLSTM^[20]	13.63	261	19.32	3 560	14.21	285	22.45	4 886
RCNN-ABiLSTM	12.98	258	19.16	2 980	13.24	246	22.29	3 795

[1]	赵庆媛, 赵志强, 叶春茂, 鲁耀兵. 基于自注意力的双波段预警雷达微动融合识别[J]. 系统工程与电子技术, 2023, 45(3): 708-716.
[2]	曹鹏宇, 杨承志, 陈泽盛, 王露, 石礼盟. 基于深度残差收缩注意力网络的雷达信号识别方法[J]. 系统工程与电子技术, 2023, 45(3): 717-725.
[3]	韩啸, 陈世文, 陈蒙, 杨锦程. 基于互易点学习的LPI信号开集识别[J]. 系统工程与电子技术, 2022, 44(9): 2752-2759.
[4]	徐平亮, 崔亚奇, 熊伟, 熊振宇, 顾祥岐. 生成式中断航迹接续关联方法[J]. 系统工程与电子技术, 2022, 44(5): 1543-1552.
[5]	孙世岩, 张钢, 梁伟阁, 佘博, 田福庆. 基于时间序列数据扩增和BLSTM的滚动轴承剩余寿命预测方法[J]. 系统工程与电子技术, 2022, 44(3): 1060-1068.
[6]	吴涛, 王伦文, 朱敬成. 基于迁移学习和注意力机制的伪装图像分割[J]. 系统工程与电子技术, 2022, 44(2): 376-384.
[7]	金涛, 王晓峰, 田润澜, 张歆东. 基于改进1DCNN+TCN的雷达辐射源快速识别方法[J]. 系统工程与电子技术, 2022, 44(2): 463-469.
[8]	马钰棠, 孙鹏, 张杰勇, 王鹏, 闫云飞, 赵亮. 样本不平衡下的空中群组意图识别方法[J]. 系统工程与电子技术, 2022, 44(12): 3747-3755.
[9]	唐一强, 杨霄鹏, 朱圣铭. 基于注意力机制的混合CNN-BiLSTM低轨卫星信道预测算法[J]. 系统工程与电子技术, 2022, 44(12): 3863-3870.
[10]	高涌荇, 王旭东, 汪玲, 朱岱寅, 郭军, 孟凡旺. 基于RCNN的双极化气象雷达天气信号检测[J]. 系统工程与电子技术, 2022, 44(11): 3380-3387.
[11]	韩亚娟, 章露露, 雷小虎. 基于改进FDOD度量的航空发动机健康状况评估[J]. 系统工程与电子技术, 2022, 44(11): 3579-3586.
[12]	曲凌志, 杨俊安, 刘辉, 黄科举. 嵌入注意力机制的通信辐射源个体识别方法[J]. 系统工程与电子技术, 2022, 44(1): 20-27.
[13]	刘紫燕, 马珊珊, 梁静, 朱明成, 袁磊. 注意力机制CNN的毫米波大规模MIMO系统信道估计算法[J]. 系统工程与电子技术, 2022, 44(1): 307-312.
[14]	陈云翔, 饶益, 蔡忠义, 王泽洲. 基于改进相似性的装备部件剩余寿命预测及经济性储备策略[J]. 系统工程与电子技术, 2021, 43(9): 2688-2696.
[15]	崔邦彦, 田润澜, 王东风, 崔钢, 石静苑. 基于注意力机制和改进CLDNN的雷达辐射源识别[J]. 系统工程与电子技术, 2021, 43(5): 1224-1231.