350 rub
Journal Neurocomputers №10 for 2014 г.
Article in number:
Diagnostics of flight personnel using flight simulators data
Keywords: markov model multifactor Markov network diagnostics flight simulator
Authors:
L.S. Kuravsky - Dr.Sc. (Eng.), Professor, Dean of Computer Science Faculty, Moscow State University of Psychology and Education. E-mail: l.s.kuravsky@gmail.com
P.A. Marmalyuk - Ph.D. (Eng.), Associate Professor, Computer Science Faculty, Moscow State University of Psychology and Education. E-mail: ykk.mail@gmail.com
G.A. Yuryev - Ph.D. (Phys.-Math.), Associate Professor, Computer Science Faculty, Moscow State University of Psychology and Education. E-mail: nezdeshni@gmail.com
S.N. Baranov - Ph.D. (Eng.), General Director, Russian Aviation Company (Moscow). E-mail: rusavia@rusavia.com
G.N. Poleshchuk - Head of Design Bureau, CSTS «Dynamika» (Moscow). E-mail: poleschuk@dinamika-avia.ru
A.A. Smirnov - Chief Engineer-Programmer, CSTS «Dynamika» (Moscow). E-mail: smirnov_a_a_@mail.ru A.N. Shishov - Leading Engineer, Pilot-Instructor, CSTS «Dynamika» (Moscow)
Abstract:
Presented is a new technology for diagnostics of flight personnel using flight simulators data, which is based on probabilistic estimations of the likelihood and multifactor Markov models to be identified. Goodness-of-fit evaluations for these models are determined with the aid of statistical tests. To solve the identification problem, a numerical procedure of multi-dimensional non-linear optimization is performed, which provides the solution of an inverse problem for the continuous-time discrete-state Markov model Kolmogorov differential equations system. As a result of its solution, a set of free parameters estimations is obtained, which determines the system of equations yielding time probability functions which approximate a sample probability distribution at the given time points. The classifier-building technique, which makes it possible to assess a degree of correspondence between observed data and reference distributions identified for different diagnosed groups, is considered. The approach in question was practically applied for diagnostics of flight personnel using helicopter flight simulator data.
Pages: 14-23
References

  1. Kim Dzh.O., M'yuller Ch.U., Klekka U.R. Faktornyy, diskriminantnyy i klasternyy analiz. M.: Finansy i statistika. 1989. 215 s.
  2. Dobeshi I. Desyat' lektsiy po veyvletam. M.: RKhD. 2001.13.       Dobeshi I. Desyat' lektsiy po veyvletam. M.: RKhD. 2001.
  3. Kramer G. Matematicheskie metody statistiki. M.: Mir. 1976. 648 s.
  4. Kuravskiy L.S., Yur'ev G.A. Veroyatnostnyy metod fil'tratsii artefaktov pri adaptivnom testirovanii // Eksperimental'naya psikhologiya, 2012. T. 5. № 1. S. 119-131.
  5. Kuravskiy L.S. Yur'ev G.A. Primenenie fil'tra Kalmana dlya fil'tratsii artefaktov pri adaptivnom testirovanii // Informatsionnye tekhnologii. 2012. №4. S.63-69.
  6. Kuravskiy L.S., Baranov S.N., Kornienko P.A. Obuchaemye mnogofaktornye seti Markova i ikh primenenie dlya issledovaniya psikhologicheskikh kharakteristik // Neyrokomp'yutery: razrabotka, primenenie. 2005. № 12. S. 65-76.
  7. Kuravskiy L.S., Baranov S.N., Yur'ev G.A. Sintez i identifikatsiya skrytykh markovskikh modeley dlya diagnostiki ustalostnogo razrusheniya // Neyrokomp'yutery: razrabotka, primenenie. 2010. № 12. S. 20-36.
  8. Kuravskiy L.S., Marmalyuk P.A., Alkhimov V.I., Yur'ev G.A. Matematicheskie osnovy novogo podkhoda k postroeniyu protsedur testirovaniya // Eksperimental'naya psikhologiya. 2012. T. 5. № 4. S. 75-98.
  9. Kuravskiy L.S., Marmalyuk P.A., Barabanshchikov V.A., Bezrukikh M.M., Demidov A.A., Ivanov V.V., Yur'ev G.A. Obuchaemye struktury kak osnova programmnogo obespecheniya dlya diagnostiki navykov i kompetentsiy // Neyrokomp'yutery: razrabotka i primenenie. 2013. № 5. S. 46-60.
  10. Kuravskiy L.S., Marmalyuk P.A., Baranov S.N., Abramochkina V.I., Petrova Ye.A. Faktornyy analiz rezul'tatov veyvlet-preobrazovaniy longityudnykh dannykh kak novyy metod issledovaniya dinamicheskikh kharakteristik slozhnykh sistem // Neyrokomp'yutery: razrabotka, primenenie. 2009. № 9. S. 5(19.
  11. Markovskie modeli v zadachakh diagnostiki i prognozirovaniya: Ucheb. posobie / Pod red. L.S. Kuravskogo. M.: RUSAVIA. 2013. 172 s.
  12. Kuravsky L.S. and Baranov S.N. The concept of multifactor Markov networks and its application to forecasting and diagnostics of technical systems // Proc. Condition Monitoring. 2005. Cambridge, United Kingdom. R. 111-117.
  13. Kuravsky L.S., Baranov S.N., Yuryev G.A. Synthesis and identification of hidden Markov models based on a novel statistical technique in condition monitoring // Modeling and data analysis. 2012. № 1. R. 4-24.
  14. Kuravsky L.S., Margolis A.A., Yuryev G.A., Marmalyuk P.A. Decision support system for testing CM personnel professional skills // Proc. 9th International Conference on Condition Monitoring & Machinery Failure Prevention Technologies, London, UK. June 2012. 8 p.