THE PROBLEM OF THE SELECTION OF A NEURAL NETWORK MODEL FOR PREDICTING THE STREAMS OF DATA OF DISTRIBUTED INFORMATION SYSTEMS

The object of research in the work are distributed information systems, at the entrance of which comes a stream of requests that require the implementation of certain resources for their service. The subject of research is the prediction of data flows in such systems. The purpose of the work is to analyze the problem of choosing a neural network model for predicting the data flows of distributed information systems. The specificity of the studied problem is analyzed, as well as approaches to the solution based on the theory of queuing systems. The conclusion is drawn about the insufficient adequacy of such systems under conditions of a dynamic change of state. In this regard, it became necessary to develop their own specialized mathematical and algorithmic apparatus. As a result, an approach was proposed to reduce the sample size on the basis of combining the neural network model with a numerical method that takes into account the known regularities of the function and exempts the neural network from predicting these regularities. The specificity of the mathematical apparatus required the use of appropriate algorithmic support for its solution. Thus, the analysis of the problem of choosing a neural network model for predicting the data flows of distributed information systems was carried out.

1. Gnedenko B.V., Kovalenko I.N. Vvedenie v teoriyu massovogo obsluzhivaniya. - M.: Nauka, 1987. – 336 p.

2. Bronshteyn O.I., Dukhovnyy I.M. Modeli prioritetnogo obsluzhivaniya v informatsionno-vychislitel'nykh sistemakh. - M.: Nauka, 1976. - 220 p.

3. Glushkov V.M., Gusev V.V., Mar'yanovich T.P., Sakhnyuk M.A. Programmnye sredstva modelirovaniya nepreryvno-diskretnykh sistem. - Kiev: Naukova dumka, 1975.

4. Ivakhnenko A.G., Krotov G.I., Cheberkus V.I. Harmonic and exponentialharmonic GMDH algorithms for long-term prediction of os-cillating processes. Part I. Sov. J. of Automation and Information Sci-ences, v.14, no.l, 1981, pp. 3-17.

5. Muller J.-A. Analysis and prediction of ecological systems. SAMS, vol.21, 1996.

6. Galushkin A.I., Tomashevich D.S., Tomashevich N.S., Muromskiy M.Yu., Shachnev E.A. Neyronnye algoritmy ekstrapolyatsii funktsiy i ikh primenenie v zadachakh prognozirovaniya raboty Call-tsentrov. Chast' 1. // Neyrokomp'yutery. - No. 2, 2000. - 12 p.

7. Zaentsev I.V. Prognozirovanie zagruzki lokal'noy vychislitel'¬noy seti peydzhingovogo tsentra na osnove neyronnykh setey// Studencheskie nauchnye soobshcheniya (vyp. 2). Tez. dokl. - Voronezh: VGU, 1998. - pp. 27.

8. Nikolis G., Prigozhiy I. Samoorganizatsiya v neravnovesnykh sistemakh. - M.: Mir, 1979. - 309 p.

9. Nikolis G. Dinamika ierarkhicheskikh sistem. - M.: Mir, 1989. - 486 p

10. Surovtsev I.S., Klyukin V.I., Pivovarova R.P. Neyronnye seti. Vvedenie v sovremennuyu informatsionnuyu tekhnologiyu. - Voronezh: VGU, 1994. - 224 p.

11. Rozenblatt F. Printsipy neyrodinamiki (Pertseptrony i teoriya mekhanizmov mozga) - M.: Energiya, 1965. - 480 p.

12. Minskiy M., Paypert S. Perseptrony. - M.: Mir, 1971. - 261 p.

13. Uossermen F. Neyrokomp'yuternaya tekhnika: teoriya i praktika. - M.: Mir, 1992. - 180 p.

14. Gorban' A. N. Obuchenie neyronnykh setey. - M.: SP Paragraf,1990. -159 p.

15. Muller В., Reinhardt J. Neural Networks. An introduction. - Berlin: Springer-Verlag, 1991. - 266p.

16. Mkrtchyan S. O. Neyrony i neyronnye seti. (Vvedenie v teoriyu formal'nykh neyronov) - M.: Energiya, 1971. - 232 p.

17. Vol'kenshteyn M.V. Biofizika: Ucheb. rukovodstvo. - M.: Nauka, Gl. red. fiz.-mat. lit., 1988. - 592 p.

18. Kuffler S., Nikole Dzh. Ot neyrona k mozgu. - M.: Mir, 1979. - 440 p.

19. Sokolov E.N., Vaytkyavichyus G. G. Neyrointellekt: ot neyrona k neyrokomp'yuteru. - M.: Nauka, 1989. - 238 p.

20. Sokolov E.N., Shmelev L. A. Neyrobionika. (Organizatsiya neyropodobnykh elementov i sistem).- M.: Nauka, 1983. - 280 p.

21. Ripley B.D. Pattern Recognition and Neural Networks. - Cambridge: Cambridge University Press, 1996. - 403 p.

22. Masters T. Signal and Image Processing with Neural Networks: A C++ Sourcebook. - New York: Wiley, 1994. - 286 p.

23. Volobuev N.A., Neganov V.A., Nefedov E.I., Romanchuk P.I. Kvantovomekhanicheskie effekty pri rabote ionnykh kanalov// Vestnik novykh meditsinskikh tekhnologiy. - 1997, No.1-2. - 16 p.

24. Lawrence S., Tsoi A.C., BackA.D. Function Approximation with Neu¬ral Networks and Local Methods: Bias, Variance and Smoothness// Australian Conference on Neural Networks, ACNN - 1996. - pp. 16-21.

25. Hornik K. Some New Results on Neural Network Approximation// Neural Networks. - 1993, No.6. - P. 1069-1072.

26. Yang Т., Chua L. Implementing Back-Propagation-Through-Time Learning Algorithm Using Cellular Neural Networks. // International Journal on Bifurcation and Chaos. - 1999, Vol. 9, No.6. - P. 1041-1074.

27. Bondarenko E.V. Self-organization Processes in Chaotic Neural Net works Under External Periodic Force // International Journal of Bifurcation and Chaos. - 1997, Vol. 7, No. 8. - P. 1887-1895.

28. Заенцев И.В. Критерии эффективности обработки информации в нейронных сетях. // Межвуз. НПК "Актуальные проблемы совершенствования научно-технического обеспечения деятельности ОВД": Воронеж. Воронежский институт МВД России, 1999. - С. 109- 111.

29. Draghici S. Sethi I.К. On the Possibilities of the Limited Precision Weights Neural Networks in Classification Problems. // Australian Conference on Neural Networks, ACNN - 1996. - P. 132-139.