Optimized age dependent clustering algorithm for prognosis: A case study on gas turbines

Document Type : Article

Authors

1 Department of Mechanical Engineering, Sharif University of Technology, Tehran, P.O. Box 11155-9567, Iran

2 Faculty of ECE, Tehran University, Tehran, P.O. Box 14395-515, Iran

10.24200/sci.2020.53863.3459

Abstract

This paper proposes an Age Dependent Clustering (ADC) structure to be used for prognostics. To achieve this aim, a step-by-step methodology is introduced, that includes clustering, reproduction, mapping and finally estimation of Remaining Useful Life (RUL). In the mapping step, neural fitting tool is used. Considering age based clustering concept, determination of main elements of the ADC model is discussed. Genetic algorithm (GA) is used to find the elements of the optimal model. Lastly, fuzzy technique is applied to modify the clustering. The efficacy of the proposed method is demonstrated with a case study on the health monitoring of some turbofan engines. The results show that the concept of clustering even without optimization processes is efficient even for the simplest form of performance. However, by optimizing structure elements and fuzzy clustering, the prognosis accuracy increases up to 71%. The effectiveness of age dependent clustering in prognosis is proven in comparison with other methods.

Keywords


References
1. Gao, H., Wang, A., Bai, G., Wei, C., and Fei, C.Substructure-based distributed collaborative probabilisticanalysis method for low-cycle fatigue damageassessment of turbine blade-disk", Aerospace Scienceand Technology, 1(79), pp. 636{46 (2018).2. Tahir, M., Aslam, M., Hussain, Z., Abid, M., andHaider Bhatti, S. Bayesian analysis of heterogeneousdoubly censored lifetime data using the 3-componentmixture of Rayleigh distributions: A Monte Carlosimulation study", Scientia Iranica, Transactions onIndustrial Engineering (E), 26, pp. 1789{1808 (2018).3. Ding, S., Wang, Z., Qiu, T., Zhang, G., Li, G.,and Zhou, Y. Probabilistic failure risk assessmentfor aeroengine disks considering a transient process",Aerospace Science and Technology, 1(78), pp. 696{707(2018).4. Amezquita-Sanchez, J.P. and Adeli, H. Feature extractionand classi cation techniques for health monitoringof structures", Scientia Iranica, TransactionsA, Civil Engineering, 22(6), pp. 1931{1940 (2015).5. Jardine, A.K., Lin, D., and Banjevic, D. A reiew onmachinery diagnostics and prognostics implementingcondition-based maintenance", Mechanical Systemsand Signal Processing, 20(7), pp. 1483{1510 (2006).6. Saxena, A., Goebel, K., Simon, D., and Eklund, N.Damage propagation modeling for aircraft enginerun-to-failure simulation", In Prognostics and HealthManagement, PHM 2008, International Conference,pp. 1{9 (2008).7. Amezquita-Sanchez, J.P. and Adeli, H. Optimal tunerselection using Kalman lter for a real-time modulargas turbine model", Transaction A, Civil Engineering,22(6), p. 1940 (2015).8. Zhao, W. A probabilistic approach for prognosticsof complex rotary machinery systems", PhD Thesis,University of Cincinnati (2015).9. Vachtsevanos, G., Wang, P., and Khiripet, N.Prognostication: algorithms and performance assessmentmethodologies", In ATP Fall National MeetingCondition-Based Maintenance Workshop, pp. 15{17,San Jose, California (1999).10. Li, L.L., Ma, D.J., and Li, Z.G. Residual usefullife estimation by a data-driven similarity-based approach",Quality and Reliability Engineering International,
33(2), pp. 231{9 (2017).11. Baraldi, P., Compare, M., Sauco, S., and Zio, E.Ensemble neural network-based particle ltering forprognostics", Mechanical Systems and Signal Processing,41(1), pp. 288{300 (2013).1256 A. Mahmoodian et al./Scientia Iranica, Transactions B: Mechanical Engineering 28 (2021) 1245{125812. Javed, K., Gouriveau, R., and Zerhouni, N. SWELM:
A summation wavelet extreme learning machinealgorithm with a priori parameter initialization", Neurocomputing,10(123), pp. 299{307 (2014).13. Xu, J., Wang, Y., and Xu, L. PHM-oriented integratedfusion prognostics for aircraft engines based onsensor data", IEEE Sensors Journal, 14(4), pp. 1124{1132 (2014).14. Yu, J. Aircraft engine health prognostics based onlogistic regression with penalization regularization andstate-space-based degradation framework", AerospaceScience and Technology, 68, pp. 345{361 (2017).15. Simon, D. A comparison of ltering approaches foraircraft engine health estimation", Aerospace Scienceand Technology, 12(4), pp. 276{84 (2008).16. Lu, F., Ju, H., and Huang, J. An improved extendedKalman lter with inequality constraints for gas turbineengine health monitoring", Aerospace Science andTechnology, 58, pp. 36{47 (2016).17. Son, J., Zhou, S., Sankavaram, C., Du, X., and Zhang,Y. Remaining useful life prediction based on noisycondition monitoring signals using constrained Kalman lter", Reliability Engineering & System Safety, 152,pp. 38{50 (2016).18. Zhou, D., Wu, Y., Gao, F., Breaz, E., Ravey, A.,and Miraoui, A. Degrad icle lter approach", IEEE Transactions onIndustry Applications, 53(4), pp. 4041{4052 (2017).19. Ahsan, S., Lemma, T.A., and Muhammad, M. Prognosisof gas turbine remaining useful life using particle lter approach", Mate-rialwissenschaft und Werksto technik, 50(3), pp. 336{345 (2019).20. Tahan, M., Tsoutsanis, E., Muhammad, M., andKarim, Z.A. Performance-based health monitoring,diagnostics and prognostics for condition-based maintenanceof gas turbines: A review", Applied Energy,15, pp. 122{144 (2017).21. Ragab, A., Yacout, S., Ouali, M.S., and Osman, H.Pattern-based prognostic methodology for onditionbasedmaintenance using selected and weighted survivalcurves", Quality and Reliability Engineering International,33(8), pp. 1753{1772 (2017).22. Losi, E., Venturini, M., and Manservigi, L. Gasturbine health state prognostics by means of Bayesianhierarchical models", Journal of Engineering for GasTurbines and Power, 1, pp. 141{148 (2019).23. Moghaddass, R. and Zuo, M.J. An integrated frameworkfor online diagnostic and prognostic health monitoringusing a multistate deterioration process", ReliabilityEngineering & System Safety, 124, pp. 92{104(2014).24. Huang, C.C. and Yuan, J. A two-stage preventivemaintenance policy for a multi-state deterioration system",Reliability Engineering & System Safety, 95(11),pp. 1255{1260 (2010).25. Soro, I.W., Nourelfath, M., and At-Kadi, D. Performanceevaluation of multi-state degraded systemswith minimal repairs and imperfect preventive maintenance",Reliability Engineering & System Safety,95(2), pp. 65{69 (2010).26. Dong, M. and He, D. Hidden semi-Markov modelbasedmethodology for multi-sensor equipment healthdiagnosis and prognosis", European Journal of OperationalResarch, 178(3), pp. 858{878 (2007).27. Nguyen, K.T., Fouladirad, M., and Grall A. Modelselection for degradation modeling and prognosis withhealth monitoring data", Reliability Engineering &System Safety, 169, pp. 10{16 (2018).28. De Giorgi, M.G., Ficarella, A., and De Carlo, L.Jet engine degradation prognostic using arti cialneural networks", Aircraft Engineering and AerospaceTechnology, 92(3), pp. 296{303 (2019).29. Lu, F., Wu, J., Huang, J., and Qiu, X. Aircraft enginedegradation prognostics based on logistic regressionand novel OS-ELM algorithm", Aerospace Science andTechnology, 84, pp. 661{671 (2019).30. M ul Hassan, M., Danish, F., Yousuf, W.B., andKhan, T.M. Comparison of di erent life distributionschemes for prediction of crack propagation in anaircraft wing", Engineering Failure Analysis, 1(96),pp. 241{254 (2019).31. Huang, H.Z., Wang, H.K., Li, Y.F., Zhang, L., and
Liu, Z. Support vector machine based estimationof remaining useful life: current research status andfuture trends", Journal of Mechanical Science andTechnology, 29(1), pp. 151{163 (2015).32. Goebel, K., Saha, B., and Saxena, A. A comparisonof three data-driven techniques for prognostics", In62nd Meeting of the Society for Machinery FailurePrevention Technology (mfpt), pp. 119{131 (2008).33. Razavi, S.A., Najafabadi, T.A., and Mahmoodian, A.A prognosis methodology based on enhanced lolimot
algorithm using historical data", In 2019 Prognosticsand System Health Management Conference (PHMParis),pp. 35{38 (2019).34. Zarandi, M.F., Faraji, M.R., and Karbasian, M. Anexponential cluster validity index for fuzzy clusteringwith crisp and fuzzy data", Scientia Iranica, TransactionsE, Industrial Engineering, 17(2), pp. 90{95(2010).35. Saxena, A. and Goebel, K. C-MAPSS data set, NASAAmes prognostics data repository", last retrieved fromhttp://ti.arc.nasa.gov/project/ prognostic-d atarepository,
NASA Ames, Mo ett Field, CA (2008).A. Mahmoodian et al./Scientia Iranica, Transactions B: Mechanical Engineering 28 (2021) 1245{1258 125736. Ramasso, E. and Saxena, A. Review and analysisof algorithmic approaches developed for prognosticson CMAPSS dataset", In Annual Conference of thePrognostics and Health Management Society, pp. 128{134 (2014).37. Rezvani, K., Maia, N.M., and Sabour, M.H. Acomparison of some methods for structural damage detection",Scientia Iranica, Transactions B, MechanicalEngineering, 25(3), pp. 1312{1322 (2018).38. Javed, K., Gouriveau, R., Zemouri, R., and Zerhouni,N. Features selection procedure for prognostics: Anapproach based on predictability", Reliability Engineering
& System Safety, 15, pp. 165{175 (2017).39. Le Son, K., Fouladirad, M., Barros, A., Levrat, E.,and Iung, B. Remaining useful life estimation basedon stochastic deterioration models: A comparativestudy", Reliability Engineering & System Safety, 12,pp. 165{175 (2013).
40. Mohammadi, E. and Montazeri-Gh, M. Simulation offull and part-load performance deterioration of industrialtwo-shaft gas turbine", Journal of Engineering forGas Turbines and Power, 136(9), pp. 602{609 (2014).41. Li, Y.G. and Nilkitsaranont, P. Gas turbine performanceprognostic for condition-based maintenance",Applied Energy, 86(10), pp. 2152{2161 (2009).42. Diallo, O.N. A data analytics approach to gas turbineprognostics and health management", Doctoral dissertation,Georgia Institute of Technology (2010).43. Su, Y., Tao, F., Jin, J., Wang, T., Wang, Q., andWang, L. Failure prognosis of complex equipmentwith multistream deep recurrent neural network",Journal of Computing and Information Science inEngineering, 20(2) (2020).44. Mahmoodian, A., Durali, M., and Saadat, M. Investigatingdi erent structures for mapping sensor informationof a complex mechanical system to its healthstatus", Proceedings of the 26th ISME Conference, pp.110{114 (2018).45. Ramasso, E. Investigating computational geometryfor failure prognostics", International Journal of Prognosticsand Health Management, 5(1), pp. 165{178(2014).46. Javed, K., Gouriveau, R., and Zerhouni, N. Novel failureprognostics approach with dynamic thresholds formachine degradation", Industrial Electronics Society,IECON 2013-39th Annual Conference of the IEEE, pp.4404{4409 (2013).47. Saxena, A., Celaya, J., Balaban, E., Goebel, K.,Saha, B., Saha, S., and Schwabacher, M. Metricsfor evaluating performance of prognostic techniquesin prognostics and health management", PHM 2008,International Conference, pp. 1{17 (2008).48. Mahmoodian, A., Durali, M., and Saadat, M. A novelprognostic model of performance degradation based onfusion of current and historical predictions (FCHP)",Proceedings of the 6th GTC Conference, pp. 210{212(2018).49. Vazirizade, M., Bakhshi, A., and Bahar, O. Onlinenonlinear structural damage detection using HilbertHuang transform and arti cial neural networks",Scientia Iranica, Transactions A, Civil Engineering,26(3), pp. 180{188 (2019).50. Zarandi, M.F., Faraji, M.R., and Karbasian, M. Anexponential cluster validity index for fuzzy clusteringwith crisp and fuzzy data", Scientia Iranica, TransactionsE, Industrial Engineering, 17(2), pp. 54{69(2010).51. Ramasso, E., Rombaut, M., and Zerhouni, N. Jointprediction of observations and states in time-series:a partially supervised prognostics approach based onbelief functions and knn. Networks", InternationalConference on PHM, pp. 11{17 (2013).52. Khelif, R., Malinowski, S., Chebel-Morello, B., andZerhouni, N. RUL prediction based on a newsimilarity-instance based approach. In Industrial Electronics(ISIE)", 2014 IEEE 23rd International Symposium,pp. 2463{2468 (2014).