文章快速检索 高级检索

Similarity-based remaining useful life prediction method under varying operational conditions
LI Qi , GAO Zhanbao , LI Shanying , LI Baoan
School of Automation Science and Electrical Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
Received: 2015-06-17; Accepted: 2015-07-17; Published online: 2016-06-25 12: 00
Foundation item: the Fundamental Research Funds for the Central Universities(YWF-14-ZDHXY-16)
Corresponding author. Tel 010-82338693 E-mail:gaozhanbao@bjtu.edu.cn.
Abstract: Remaining useful life (RUL) prediction is the core task of prognostic and health management (PHM). A similarity-based RUL prediction method under varying operational conditions was presented. Similarity-based RUL prediction method does not need to build a model for entire complex system but can provide reasonable results, which is promising in engineering practice. However, the operational conditions such as different working loads and environmental conditions are not considered for degradation modeling. Therefore, this method combines basic similarity-based method and the effect of operational conditions to achieve better RUL prediction accuracy. Degradation models with different operational conditions were built by training units, and the RUL prediction was achieved by matching corresponding model using the real-time operational conditions of the running unit. The proposed degradation models describe the degradation process more precisely by taking the differences of operational conditions into account. According to the same accuracy standard, multi-group numerical experiments were finished by basic similarity-based method and the proposed method. The result shows the proposed method has a higher accuracy in RUL prediction.
Key words: remaining useful life (RUL)     prognostics     operational condition     similarity     health indicator

1 算法框架

 图 1 变工况下基于相似性的RUL预测方法框架 Fig. 1 Framework of similarity-based RUL estimation method under varying operational conditions
2 算法原理 2.1 特征提取

2.2 退化模型建立

 (1)

 图 2 单工况和多工况条件下的设备退化曲线 Fig. 2 Device degradation curves under single operational condition and mutilple operational conditions
 (2)

2.3 RUL预测

 图 3 变工况下RUL预测方法滑动评估过程 Fig. 3 Moving estimation process of RUL under varying operational conditions

 (3)

 (4)

 (5)

 (6)

 (7)
3 算例分析

3.1 2008 PHM Data Challenge数据介绍

 图 4 6种不同工况的划分 Fig. 4 Six different operational conditions partition
 图 5 218个训练单元的第11号传感器测量数据 Fig. 5 Sensor measurements of sensor 11 for 218 training units

 (8)

3.2 模型训练过程

 (9)

 (10)

3.3 剩余寿命预测过程

 图 6 候选集参数θ对RUL预测性能的影响 Fig. 6 Influence of parameter θ on RUL estimation performance from candidate set

 (11)
3.4 结果评估与讨论

 数量 最优传感器组合 score 对比 MSE对比 原方法 本方法 原方法 本方法 2 4,15 1 636 1 576(↓4%) 334 314(↓6%) 3 2,3,4 2 729 1 515(↓44%) 396 311(↓21%) 4 2,3,4,11 1 545 1 173(↓24%) 305 280(↓8%) 5 2,3,4,11,15 1 043 1 059(↑1%) 258 241(↓7%) 6 2,3,4,11,15,21 988 889(↓10%) 253 229(↓9%) 7 2,3,4,11,12,15,21 1 012 856(↓15%) 259 226(↓13%) 全集 2,3,4,7,11,12,15,21 1 032 1 007(↓3%) 263 241(↓8%) 原组合 2,3,4,7,11,12,15 1 049 1 011(↓4%) 276 250(↓9%)

4 结 论

1) 提出了一种变工况条件下的基于相似性的剩余寿命预测方法,可以将工况信息纳入到退化建模中以提高模型精度。

2) 算法可实现较优的剩余使用寿命预测,例如算例中,不同传感器组合下,本方法score指标平均降低13%,MSE指标平均降低10%。

3) 本方法对处于不同工况变化规律的服役样本能够实现满足其工况特性的差异化的预测,使预测更接近工程实践。

4) 在算例分析中,验证了相似性方法中预测评估候选集的选择和传感器组合的优化对预测性能的显著影响,这对方法进一步的深入研究提供了指导。

 [1] DOYEN L, GAUDOIN O. Modeling and assessment of aging and efficiency of corrective and planned preventive maintenance[J]. IEEE Transactions on Reliability,2011, 60 (4) : 759 –769. Click to display the text [2] HENG A, ZHANG S, TAN A C, et al. Rotating machinery prognostics: State of the art,challenge and opportunities[J]. Mechanical Systems and Signal Processing,2009, 23 (3) : 724 –739. Click to display the text [3] TANG D, MAKIS V, JAFARI L, et al. Optimal maintenance policy and residual life estimation for a slowly degrading system subject to condition monitoring[J]. Reliability Engineering & System Safety,2015, 134 : 198 –207. Click to display the text [4] VACHTSEVANOS G, LEWIS F, ROEMER M, et al. Intelligent fault diagnosis and prognosis for engineering systems[M]. New Jersey: Wiley, 2006 : 289 -300. Click to display the text [5] LIAO L X, FELIX K. Review of hybrid prognostics approaches for remaining useful life prediction of engineered systems,and an application to battery life prediction[J]. IEEE Transactions on Reliability,2014, 63 (1) : 191 –207. Click to display the text [6] SI X S, WANG W, HU C H, et al. Remaining useful life estimation:A review on the statistical data driven approaches[J]. European Journal of Operational Research,2011, 213 (1) : 1 –14. Click to display the text [7] WANG T Y.Trajectory similarity based prediction for remaining useful life estimation[D].Cincinnati:University of Cincinnati,2010:39-56. Click to display the text [8] BIAN L, GEBRAEEL N, KHAROUFEH J P. Degradation modeling for real-time estimation of residual lifetimes in dynamic environments[J]. ⅡE Transactions,2015, 47 (5) : 471 –486. Click to display the text [9] YOU M, MENG G. A Framework of similarity-based residual life prediction approaches using degradation histories with failure,preventive maintenance and suspension events[J]. IEEE Transactions on Reliability,2013, 62 (1) : 127 –135. Click to display the text [10] ZHANG Q, TSE P, WAN X, et al. Remaining useful life estimation for mechanical systems based on similarity of phase space trajectory[J]. Expert Systems with Applications,2015, 42 (5) : 2353 –2360. Click to display the text [11] YOU M, MENG G. Toward effective utilization of similarity based residual life prediction methods:Weight allocation,prediction robustness,and prediction uncertainty[J]. Journal of Process Mechanical Engineering,2013, 227 (1) : 74 –84. Click to display the text [12] SAXENA A,GOEBEL K.Turbofan engine degradation simulation dataset[EB/OL].Washington,D.CNASA Ames Research Center,2008(2013-09-12).http://ti.arc.nasa.gov/project/prong-ostic-data-repository. Click to display the text [13] WANG T Y,YU J B,SIEGEL D,et al.A similarity-based prognostics approach for remaining useful life estimation of engineered systems[C]//2008 International Conference on Prognostics and Health Management(PHM 2008). Piscataway,NJ:IEEE Press,2008:1-6. Click to display the text [14] 高占宝, 李行善, 梁旭, 等. 工程系统健康描述及基于GFRF方法的健康监测[J]. 北京航空航天大学学报,2006, 32 (9) : 1026 –1030. GAO Z B, LI X S, LIANG X, et al. Engineering system health formulation and health monitoring based on GFRF approach[J]. Journal of Beijing University of Aeronautics and Astronautics,2006, 32 (9) : 1026 –1030. (in Chinese). Cited By in Cnki (0) | Click to display the text [15] LE S K, FOULADIRAD M, BARROS A, et al. Remaining useful life estimation based on stochastic deterioration models:A comparative study[J]. Reliability Engineering & System Safety,2013, 112 (4) : 165 –175. Click to display the text [16] XU J, WANG Y, XU L. PHM-oriented integrated fusion prognostics for aircraft engines based on sensor data[J]. IEEE Sensor Journal,2014, 14 (4) : 1124 –1132. Click to display the text [17] HU C, YOU B, WANG P, et al. Ensemble of data-driven prognostic algorithms for robust prediction of remaining useful life[J]. Reliability Engineering & System Safety,2012, 103 (3) : 120 –135. Click to display the text

#### 文章信息

LI Qi, GAO Zhanbao, LI Shanying, LI Baoan

Similarity-based remaining useful life prediction method under varying operational conditions

Journal of Beijing University of Aeronautics and Astronsutics, 2016, 42(6): 1236-1243
http://dx.doi.org/10.13700/j.bh.1001-5965.2015.0396