Abstract
Machine learning-based predictive modeling is to develop machine learning-based or data-driven models to predict failures before they occur and estimate the remaining useful life or time to failure (TTF) accurately. Accurate TTF estimation plays a vital role in predictive maintenance or PHM (Prognostic and Health Management). Despite the availability of large amounts of data and a variety of powerful data analysis methods, predictive models developed for PHM still fail to provide accurate and precise TTF estimations. This paper addresses this problem by integrating machine learning algorithms such as classification, regression and clustering. A classification system is used to determine the likelihood of component failures such that rough indications of TTF are provided. Clustering and SVM-based local regression are then introduced to refine the time to failure estimations provided by the classification system. The paper illustrates the applicability of the proposed approach through a real world aerospace application with details on data pre-processing requirements. The results demonstrate that the proposed method can reduce uncertainty in estimating time to failure, which in turn helps augment the usefulness of predictive maintenance.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.References
Liu K, Gebraeel NZ, Shi J (2013) A data-level fusion model for developing composite health indices for degradation modeling and prognostic analysis. IEEE Trans Autom Sci Eng 10(3):652–664
You MY, Li L, Meng G, Ni J (2010) Cost-effective updated sequential predictive maintenance policy for continuously monitored degrading systems. IEEE Trans Autom Sci Eng 7(2):581–265
Compare M, Zio E (2014) Predictive maintenance by risk sensitive particle filtering. IEEE Trans Reliab 63(1):134–143
Grall A, et al. (2002) Continuous time predictive maintenance scheduling for a deteriorating system. IEEE Trans Reliab 51(2):141–150
Feng L, Wang H, Si X, Zou H (2013) A State-Space-Based prognostic model for hidden and Age-Dependent nonlinear degradation process. IEEE Trans Autom Sci Eng 10(4):1072–1085
Schouten FA, Wartenhorst P Time to failure, time to repair and availability of a two unit standby system with markovian degrading units. Technical report BS-R9007, Center for Mathematics and Computer
Gebraeel N, Lawley M, Liu R, Parmeshwaran V Residual life prediction from vibration-based degradation signals: a neural network approach. IEEE Trans Ind Electron 51(3)
Grottke M, Trivedi K (2005) On a method for mending time to failure distributions. In: Proceedings of international conference on dependable systems and networks, Los Alamitos, USA, pp 560–569
Promellec B, Riant I, Tessier-Lescourret C (2006) Precise life-time prediction using demarcation energy approximation for distributed activation energy reaction. J Phys Condens Matter 18(2006):2199–2216
Sornette D, Andersen JV (2006) Optimal prediction of time-to-failure from information revealed by damage. Europhys Lett 74(5):778–784
L’etourneau S, Yang C, Drummond C, Scarlett E, Valds J, Zaluski M (2005) A domain independent data mining methodology for prognostics. In: Proceedings of the 59th meeting of the society for machine failure prevention technology, MFPT ’05, Virginia beach, Virginia, USA
L’etourneau S, Yang C, Liu Z (2008) Improving preciseness of time to failure predictions: application to APU starter. In: Proceedings of the 1st international conference on prognostics and health management, Denver, Colorado, USA
Yang C, L’etourneau S (2005) Learning to predict train wheel failures. In: Proceedings of the 11th ACM SIGKDD international conference on knowledge discovery and data mining (KDD2005), Chicago, USA, pp 516–525
L’etourneau S, Famili AF, Matwin S (1999) Data mining for prediction of aircraft component replacement. IEEE Intelligent Systems and their Applications 14(6):59–66
Hall M (2000) Correlation-based feature selection for discrete and numeric class machine learning. In: Proceedings of the 17th international conference on machine learning, A, pp 359–366
Kira K, Rendell L (1992) A practical approach to feature selection. In: Proceedings of the 9th international conference on machine learning, pp 249–256
Dietterich T (2000) An experimental comparison of three methods for constructing ensembles of decision trees: bagging, boosting and randomization. IEEE Intelligent Systems and their Applications 40:139–158
Tsoumakas IKG, Blahavas I (2004) Effective voting of heterogeneous classifiers. In: Proceedings of the 15th european conference on machine learning (MCML2004), pp 465–476
Dzeroski S, Zenko B (2002) Stacking with multi-response model trees. In: The proceeding of international workshop in multiple classifier systems (MCS2002), pp 201–211
Witten IH, Frank E (1999) Data mining: practical machine learning tools and techniques with java implementations. Morgan Kaufmann
Dzeroski S, Zenko B (2004) Is combining classifiers with stacking better than selecting the best one?. Mach Learn 54:254– 273
Compare M, Zio E (2014) Predictive maintenance by risk sensitive particle filtering. IEEE Trans Reliab 63(1):134–143
You MY, Liu F, Wang W, Meng G (2010) Statistically planned and individually monitored predictive maintenance management for continuously monitored degrading systems. IEEE Trans Reliab 59(4):744–753
Park KS (1993) Condition-based predictive maintenance by multiple logistic function. IEEE Trans Reliab 42 (4):556–560
Zaluski M, Létourneau S, Bird J, Yang C (2011) Developing data mining-based prognostic models for Cf-18 aircraft. Journal of Engine Gas Turbines Power 133(10)
Zaluski M, Létourneau S, Yang C (2009) Data Mining-based Prognostics for CF18 Aircraft Components. In: 13th CASI Aeronautics Conference, Kanata, Canada, May 5–7
Leblanca M, Tibshiranib R (1996) Combining estimates in regression and classification. J Am Stat Assoc 91(436):1641– 1650
Acknowledgments
Many people at the National Research Council of Canada have contributed to this work. We would also like to thank Air Canada for providing the data used in this research. This work is supported by the Natural Science Foundation (Grant No.61463031).
Author information
Authors and Affiliations
Corresponding author
Appendices
Acronym
- ACARS:
-
Aircraft Communications Addressing & Reporting System
- APU:
-
Auxiliary Power Unit
- EM:
-
Expectation Maximization Algorithm for clustering
- J48:
-
Decision Tree Classification Algorithm
- MSE:
-
Mean Squared Error
- NN:
-
Neural Network
- PHM:
-
Prognostic and Health Management
- SMO:
-
Sequential Minimal Optimization
- SVM:
-
Support Vector Machine
- TTF:
-
Time to Failure
- WEKA:
-
Waikato Environment for Knowledge Analysis
Notation
- p :
-
number of positive predictions
- N :
-
total number of positives made by classifiers in training dataset
- score i :
-
score from the reward function for the i th instance classified as positive
- NbrDetected :
-
number of detected failures
- NbrofCase :
-
total number of failures
- M :
-
number of observations kept before each failure
- N :
-
number of observations kept after each failure
- Sign :
-
sign of \({\sum }_{i=1}^{p} {score_{i} } \)
- TTF C :
-
TTF estimate from the classifier
- TTF R :
-
TTF estimate from the regression model
- RemainingOPH :
-
remaining operational life of a component (in hours)
- X i j :
-
The system state observation: the j th instance in i th time-series.
Rights and permissions
About this article
Cite this article
Yang, C., Letourneau, S., Liu, J. et al. Machine learning-based methods for TTF estimation with application to APU prognostics. Appl Intell 46, 227–239 (2017). https://doi.org/10.1007/s10489-016-0829-4
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10489-016-0829-4