Verification of the method for classifying the technical state of a turboshaft engine fuel regulator in the space of operational process parameters under factory test conditions

Ihor Ohanian, Sergiy Yepifanov

Abstract


The article presents the results of verifying the method for classifying the technical state of a helicopter turboshaft engine fuel regulator in the parameter space of the operational process under real factory-testing conditions. The applied diagnostic method is based on a mathematical model of the fuel regulator, considering the impact of deviations in structural parameters on diagnostic features. In this study, structural parameters subject to diagnostics were identified for each test, and parameters with high correlation coefficients in terms of their impact on diagnostic features were excluded. The necessity of the mathematical model identification of the regulator for applying the technical state classification method was substantiated. A method of single-factor identification of the fuel regulator's mathematical model, based on the working-fluid pressure measurements before the equivalent resistance of the first nozzle circuit, is described. The procedure for classifying the technical state was conducted in two stages: determining the conditional probabilities of the current regulator state (S) belonging to one of the modeled defects using Bayes' theorem, and establishing a diagnosis based on the maximum value of these probabilities. The study was conducted at JSC "FED" on seven hydromechanical fuel regulators. During the tests, the primary malfunctions were identified based on controlled parameters, and the states of the regulators were classified using the proposed method. The effectiveness of the method was analyzed according to quantitative classification quality indicators: the overall accuracy ranged from 90.48% to 100%, and micro-averaged precision ranged from 66.67% to 100%, depending on the specific test. It was determined that the method demonstrates high efficiency in most operational modes; however, its accuracy depends on the degree of correlation between the diagnosed structural parameters and the precision of the mathematical model identification. Recommendations for further improvement of the method include using expert systems and developing effective model identification algorithms. The method can be recommended for integration into production technologies for condition-based maintenance to improve diagnostic accuracy and optimize repair activities.

Keywords


fuel regulator; technical state; diagnostic method; classification; fault localization; defect; experiment; classification quality; overall classification accuracy; micro-averaged classification precision

References


Ohanian, I. V. Matematychna model palyvnoho nasosa-rehuliatora turbovalnoho dvyhuna vertolota [Mathematical model of fuel pump-regulator of helicopter turbo-shaft engine]. Aviacijno-kosmicna tehnika i tehnologia – Aerospace technic and technology, 2020, no. 7 (167), pp. 105–112. DOI: 10.32620/aktt.2020.7.15. (in Ukrainian).

Ohanian, I. V., & Yepifanov, S. Alhorytm klasyfikatsii tekhnichnoho stanu palyvnoho rehuliatora u prostori parametriv robochoho protsesu [Algorithm for classifying the technical stage of the firing regulator within the scope of operating process parameters]. Aviacijno-kosmicna tehnika i tehnologia – Aerospace technic and technology, 2021, no. 4 (173), pp. 158–169. DOI: 10.32620/aktt.2021.4sup1.22. (in Ukrainian).

Balayla, J. Bayesian updating and sequential testing: overcoming inferential limitations of screening tests. BMC Medical Informatics and Decision Making, 2022, vol. 22, iss. 6. DOI: 10.1186/s12911-021-01738-w.

Holman, J. P. Experimental Methods for Engineers. McGraw-Hill Series in Mechanical Engineering, 2012. 761 p.

Ilchenko, B. S. Diahnostuvannia funktsionalno-tekhnichnoho stanu hazo-perekachuvalnykh arehativ: monohrafiia [Diagnosis of the functional and technical plant of gas pumping units: monograph]. Kharkiv, Khark. nats. akad. misk. hosp-va. KhNAMH, 2011. 228 p. (in Ukrainian).

Yepifanov, S. V., & et al. Sintez sistem upravleniya i diagnostirovaniya gazoturbinnyx dvigatelej [Synthesis of control systems and diagnostics of gas turbine engines]. Kyiv, Publ. «Tehnika», 1998. 311 p. (in Russian).

Johnson, R. A., Wichern, D. W. Applied Multivariate Statistical Analysis. Prentice-Hall, 2002. 788 p.

Yakushenko, O. S., Korolov, P. V., Miltsov, V. Ye., Chumak, O. I., & Okhmakevych V. M. Identyfikatsiya matematychnoyi modeli aviatsiynoho HTD za polʹotnymy danymy [Identification of aviation gas turbine engine mathematical model by operational data]. Vestnik dvigatelestroyeniya - Bulletin of Engine Building, 2014, no. 2, pp. 130-138. (in Ukrainian).

Jiang, X., & Mahadevan, S. Bayesian cross-entropy methodology for optimal design of validation experiments. Measurement Science and Technology, 2006, vol. 17, iss. 7, pp. 1895–1908. DOI: 10.1088/0957-0233/17/7/031.

Ballabio, D., Grisoni, F., & Todeschini, R. Multivariate comparison of classification performance measures. Chemometrics and Intelligent Laboratory Systems, 2018, vol. 174, pp. 33-44. DOI: 10.1016/j.chemolab.2017.12.004.

Takahashi, K., Yamamoto, K., Kuchiba, A., & Koyama, T. Confidence interval for micro-averaged F1 and macro-averaged F1 scores. Applied Intelligence, 2022, vol. 52, pp. 4961-4972. DOI: 10.1007/s10489-021-02635-5.




DOI: https://doi.org/10.32620/aktt.2025.2.01