Industrial gas turbine operating parameters monitoring and data-driven prediction

• Autorzy: Pawełczyk Maciej , Fulara Szymon , Sepe Marzia , De Luca Alessandro , Badora Maciej

Identyfikatory

DOI

10.17531/ein.2020.3.2

Warianty tytułu

PL

Monitorowanie oraz bazująca na danych predykcja parametrów roboczych przemysłowej turbiny gazowej

• Języki publikacji: EN

Abstrakty

EN

The article reviews traditional and modern methods for prediction of gas turbine operating characteristics and its potential failures. Moreover, a comparison of Machine Learning based prediction models, including Artificial Neural Networks (ANN), is presented. The research focuses on High Pressure Compressor (HPC) recoup pressure level of 4th generation LM2500 gas generator (LM2500+G4) coupled with a 2-stage High Speed Power Turbine Module. The researched parameter is adjustable and may be used to balance net axial loads exerted on thrust bearing to ensure stable gas turbine operation, but its direct measurement is technically difficult implicating the need to indirect measurement via set of other gas turbine sensors. Input data for the research have been obtained from BHGE manufactured and monitored gas turbines and consists of real-time data extracted from industrial installations. Machine learning models trained using the data show less than 1% Mean Absolute Percentage Error (MAPE) as obtained with the use of Random Forest and Gradient Boosting Regression models. Multilayer Perceptron Artificial Neural Networks (MLP ANN) models are reviewed, and their performance checks inferior to Random Forest algorithm-based model. The importance of hyperparameter tuning and feature engineering is discussed.

PL

W artykule przedstawiono przegląd klasycznych i aktualnych metod przewidywania parametrów operacyjnych oraz potencjalnych usterek turbin gazowych. Dodatkowo zaprezentowano porównanie wybranych modeli opartych o uczenie maszynowe, w tym modeli wykorzystujące sztuczne sieci neuronowe. Przeprowadzone badania dotyczyły analiz poziomu ciśnienia ze sprężarki turbiny gazowej LM2500 czwartej generacji (LM2500+G4) połączonej z dwustopniową turbiną roboczą. Badany parametr podlega sterowaniu i może posłużyć do wyrównania sił osiowych działających na łożysko główne wału wysokiego ciśnienia w celu zapewnienia stabilnej i niezawodnej pracy turbiny gazowej. Jednocześnie jego bezpośredni pomiar jest kosztowny stąd potrzeba dokonania pośredniego pomiaru z wykorzystaniem innych czujników zamontowanych na turbinie. Dane wejściowe do analiz otrzymano dzięki uprzejmości producenta turbin, firmy BHGE. Zawierają one parametry bezpośrednio pobrane z monitorowanych turbin gazowych. Modele uczenia maszynowego otrzymane w wyniku analizy charakteryzują się średnim błędem procentowym (MAPE) na poziomie poniżej 1%. Najmniejszym błędem charakteryzują się modele otrzymane przy zastosowaniu lasów losowych (Random Forest) oraz gradientowego wzmacniania regresji (Gradient Boosted Regression). Przetestowano także zastosowanie wielowarstwowych, w pełni połączonych sztucznych sieci neuronowych, których efektywność okazała się niższa od modelu opartego o algorytm lasów losowych. W podsumowaniu podkreślono wagę dostosowywania hiperparametrów i inżynierii cech.

Słowa kluczowe

EN

gas turbine; machine learning; data-driven prediction; HP recoup pressure analysis

PL

turbina gazowa; uczenie maszynowe; predykcja bazująca na danych; analiza pomiarów ciśnienia sprężarki

• Wydawca: Polskie Naukowo-Techniczne Towarzystwo Eksploatacyjne
• Czasopismo: Eksploatacja i Niezawodność
• Rocznik: 2020
• Tom: Vol. 22, no. 3
• Strony: 391--399
• Opis fizyczny: Bibliogr. 38 poz., rys., tab.

Twórcy

autor

Pawełczyk Maciej

• Łukasiewicz Research Network – Institute of Aviation al. Krakowska 110/114 02-256 Warsaw, Poland

autor

Fulara Szymon

• Łukasiewicz Research Network – Institute of Aviation al. Krakowska 110/114 02-256 Warsaw, Poland

autor

Sepe Marzia

• Baker Hughes Via Felice Matteucci 2 50127 Firenze, Italy

autor

De Luca Alessandro

• Baker Hughes Via Felice Matteucci 2 50127 Firenze, Italy

autor

Badora Maciej

• Baker Hughes al. Krakowska 110/114 02-256 Warsaw, Poland

Bibliografia

• 1. Augustyn S. Energy Model of Change in Technical Condition of Aircraft Power Plants and Space Propulsion Systems. Aviation Advances & Maintenance 2017; 40(2), https://doi.org/10.1515/afit-2017-0009.
• 2. Badeer G. H. Engine Thrust Bearing Condition Monitoring Method. US Patent US no. 6,637,932 B2, GE Schenectady, NY, US, Oct. 23, 2003.
• 3. Badeer G. H. GE Aeroderivative Gas Turbines - Design and Operating Features. GE Power Systems GER-3695E (10/00) available online:https://www.ge.com/content/dam/gepower-pgdp/global/en_US/documents/technical/ger/ger-3695e-ge-aero-gas-turbine-design-op-features.pdf
• 4. Batalha E. Aircraft Engines Maintenance Costs and Reliability. An Appraisal of the Decision Process to Remove an Engine for a Shop Visit Aiming at Minimum Maintenance Unit Cost 2012.
• 5. BHGE aeroderivative gas turbines portfolio available online at https://www.bhge.com/aeroderivative-gas-turbines
• 6. Carlevaro F, Cioncolini S, Sepe M, Parrella I, Escobedo E, Allegorico C, De Stefanis L, Mastroianni M. Use of operating parameters, digital replicas and models for condition monitoring and improved equipment health. ASME Turbo Expo 2018, https://doi.org/10.1115/GT2018-76849.
• 7. Chmielewski M, Fulara S, Gieras M. Numerical Prediction of GTD-350 Turboshaft Engine Combustor Deterioration. Journal of KONES 2017; 24(2): 47-58.
• 8. Cyrus B Meher-Homji C. B, Yates D, Weyermann H. P. Aeroderivative Gas Turbine Drivers for The ConocoPhillips Optimized CascadeSM LNG Process - World's First Application and Future Potential. 15th International Conference & Exhibition on Liquefied Natural Gas 2007 available online: http://www.ivt.ntnu.no/ept/fag/tep4215/innhold/LNG%20Conferences/2007/fscommand/PS2_6_Meher_Homji_s.pdf
• 9. Deloux E, Castanier B, Berenguer C. Predictive maintenance policy for a gradually deteriorating system subject to stress. Reliability Engineering and System Safety 2009; 94: 418- 431, https://doi.org/10.1016/j.ress.2008.04.002.
• 10. Demgne J, Mercier S, Lair W, Lonchampt J. Modelling and numerical assessment of a maintenance strategy with stock through Piecewise Deterministic Markov Processes and Quasi Monte Carlo methods. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability 2017; 231(4): 429-445, https://doi.org/10.1177/1748006X17712121.
• 11. Fulara S, Chmielewski M, Gieras M. Experimental research of the small gas turbine with variable area nozzle. Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering 2019; 233(15): 5650-5659, https://doi.org/10.1177/0954410019853977.
• 12. Galar D, Gustafson A, Tormos B, Berges L. Maintenance Decision Making Based on Different Types of Data Fusion. Eksploatacja i Niezawodnosc - Maintenance and Reliability 2012; 14 (2): 135-144.
• 13. Garcia Nieto P.J, Garcia-Gonzales E, Sanches Lasheras F, de Cos Juez F.J. Hybrid PSO-SVM-based method for forecasting of the remaining useful life for aircraft engines and evaluation of its reliability. Reliability Engineering & System Safety 2015; 138: 219-231, https://doi.org/10.1016/j.ress.2015.02.001.
• 14. GE Power & Water Distributed Power. LM2500+ G4 BaseGas Turbine Fact Sheet available online: https://www.ge.com/content/dam/gepower-pgdp/global/en_US/documents/product/lm2500-plus-g4-fact-sheet.pdf accessed on September 2018
• 15. Heaton J. Automated Feature Engineering for Deep Neural Networks with Genetic Programming. College of Engineering and Computing Nova Southeastern University 2017.
• 16. Herbert L. Designing for Reliability, Maintainability, and Sustainability (RM&S) in Military Jet Fighter Aircraft Engines, Massachusetts Institute of Technology 2002.
• 17. Huang B, Wang Z. The Role of Data Prefiltering For Integrated Identification and Model Predictive Control. IFAC Proceedings Volumes 1999; 32(2): 6751-6756, https://doi.org/10.1016/S1474-6670(17)57153-0.
• 18. Huang H.Z, Tong X, Zuo M. J. Posbist fault tree analysis of coherent systems. Reliability Engineering & System Safety 2004; 84(2): 141-148, https://doi.org/10.1016/j.ress.2003.11.002.
• 19. Institute of Aviation. GE Challenge available online: https://ilot.edu.pl/tag/ge-challenge accessed on August 2018
• 20. Jankowski A, Kowalski M. Creating Mechanisms of Toxic Substances Emission of Combustion Engines. Journal of KONBiN 2015; 36(1):33-42, https://doi.org/10.1515/jok-2015-0054.
• 21. Kopytov E, Labendik V, Yunusov S, Tarasov A. Managing and Control of Aircraft Power Using Artificial Neural Networks. Proceeding of the 7th International Conference "Reliability and Statistics in Transportation and Communication 2007.
• 22. Kozik P, Sęp J. Aircraft Engine overhaul demand forecasting using ANN. Management and Production Engineering Review 2012; 3(2): 21-26.
• 23. Kumar U. D, Crocker J, Knezevic J. Evolutionary Maintenance for Aircraft Engines. Proceedings Annual Reliability and Maintainability Symposium 1999: 62-68.
• 24. Liu D, Zhang H, Polycarpou M, Alippi C, He H. Elman-Style Process Neural Network with Application to Aircraft Engine Health Condition Monitoring. Advances in Neural Networks. Proceedings of the 8th International Symposium on Neural Networks 2011.
• 25. Louppe G. Understanding Random Forests: From Theory to Practice. ArXiv Preprint 2014; ArXiv:1407.7502: 35.
• 26. Lu J-M, Innal F, Wu X-Y, Liu Y, Lundteigen M. A. Two-terminal Reliability Analysis for Multi-Phase Communication Networks. Eksploatacja i Niezawodnosc - Maintenance and Reliability 2016; 18(3): 418-427, https://doi.org/10.17531/ein.2016.3.14.
• 27. Lu J-M, Lundteigen M. A, Liu Y, Wu X-Y. Flexible Truncation Method for The Reliability Assessment of Phased Mission Systems with Repairable Components. Eksploatacja i Niezawodnosc - Maintenance and Reliability 2016; 18 (2): 229-236, https://doi.org/10.17531/ein.2016.2.10.
• 28. Michelassi V, Allegorico C, Cioncolini S, Graziano A, Tognarelli L, Sepe M. Machine learning in gas turbines, from component design to asset management. ASME Journal: Global Gas Turbine News 2018; 140(09): 54-55,https://doi.org/10.1115/1.2018-SEP5.
• 29. Mingazov B. G, Korobitsin N. A. Use of Probability Indices for Assessment of Gas Turbine Power Station Reliability under Commercial Operation Conditions. Russian Aeronautics 2009; 53(2): 226-229, https://doi.org/10.3103/S1068799810020170.
• 30. Morris A.S, Langari R. Measurement and Instrumentation: Theory and Application 2016, https://doi.org/10.1016/B978-0-12-800884-3.00017-4.
• 31. Pawełczyk M, Bibik P. Wykorzystanie nowoczesnych narzędzi inżynierskich w projektowaniu bezzałogowego wiropłata czterowirnikowego. Materiały IX Krajowego Forum Wiropłatowego. Instytut Lotnictwa 2013, https://doi.org/10.5604/05096669.1106572.
• 32. Simon D.L. An Integrated Architecture for On-Board Aircraft Engine Performance Trend Monitoring and Gas Path Fault Diagnostics. 57th Joint Army-Navy-NASA-Air Force (JANNAF) Propulsion Meeting sponsored by the JANNAF Interagency Propulsion Committee 2010.
• 33. Soumitra P, Kapoor K, Jasani D, Dudhwewala R, Gowda V. B, Gopalakrishnan Nair T.R. Application of Artificial Neural Networks in Aircraft Maintenance. Repair and Overhaul Solutions, Analysis and Manufacturing Technologies 2008.
• 34. Tibshirani R, Friedman J, Hastie T. Regularization Paths for Generalized Linear Models via Coordinate Descent. Journal of Statistical Software. 2010; 33(1), https://doi.org/10.18637/jss.v033.i01.
• 35. Verma M, Kumar A. A novel general approach to evaluating the reliability of gas turbine system. Engineering Applications of Artificial Intelligence 2014; 28: 13-21, https://doi.org/10.1016/j.engappai.2013.10.001.
• 36. Walsh P, Fletcher P. Gas Turbine Performance. Blackwell Science 2004, https://doi.org/10.1002/9780470774533.
• 37. Zadeh L.A. Fuzzy sets. Information and Control 1965; 8(3): 338-353, https://doi.org/10.1016/S0019-9958(65)90241-X.
• 38. Zhang H, Keerthi S, Mahajan D, Dhillon I.S, Hsieh C.J. Gradient Boosted Decision Trees for High Dimensional Sparse Output. In ICML;