Nguyen Quang Thuong - Dr. Sc. (Eng.), Professor, State University of Management (Moscow). E-mail: tikhonovrus@gmail.com Nguyen Van Thang - Post-graduate Student, Bauman Moscow State Technical University. E-mail: emailvictoriousvn@yahoo.com V.V. Koryanov - Ph. D. (Eng.), Associate Professor, Bauman Moscow State Technical University

During the flight tests of modern unmanned aerial vehicles (UAVs), an important issue is the analysis of the design and the functional connections of the aerodynamic model. According to the results of this analysis are corrected mathematical models of aerodynamic coefficients UAVs refined control system settings and make changes in a constructive image of the UAV, if necessary. 1. The problem of statistical synthesis models of complex systems. The problem of statistical synthesis models of complex systems is to find a model of optimal complexity, in which the received external criterion reaches its lower limit. The object of the selection is not only a mathematical model of the system under study, but still some way varies statistical sample. 2. The statistical sampling and basic functions for design and functional relationships model. The statistical sampling and various combinations of basis functions form an appropriate design and the functional connections of the model. 3. External criteria for statistical synthesis model. Regularity criterion is recommended for short-term forecasts of one - two steps forward. The criterion of the minimum offset allows you to select the model, the least sensitive to changes in the set of experimental points on which it is received. Criterion of convergence of integration process in finite models is the standard error of numerical integration on the interpolation section (ie, where the experimental points are given). It does not require separation of statistics on training and test sequence. The criterion is generally used in predictive models of autoregressive type. The balance criterion variables allows us to reduce the problem of extrapolation to the interpolation problem. 4. Operations conversion statistical sampling models. Reduction - operation information n dimensional statistical sampling to the n dimensional, which allows to reduce the original multi-dimensional problem to a sequence of one-dimensional, which is formed by n polynomials of one variable. Inversion - forming the inverse sampling operation by which to form an inverse functional relationship. This operation allows you to not only analytically the inverse function, but in the multidimensional case to implement a reduction of the original n dimensional functions to a sequence of n dimensional inverse functions. 5. The model of structural-parametric approximation of the experimental data for the synthesis of model systems. The model of structural-parametric approximation is constructed based on the requirement that all known approximating polynomial (power series, fractional rational functions, Fourier series, etc.) is a special case of this model.

 

1. Ivakhnenko A.G. Induktivnyjj metod samoorganizacii modelejj slozhnykh sistem. Kiev: Tekhnika. 1981.
2. Ivakhnenko A.G. Samoorganizacija prognozirujushhikh modelejj. Kiev:Tekhnika. 1985. 223 s.
3. Balyk V.M. Statisticheskijj sintez proektnykh reshenijj pri razrabotke slozhnykh sistem. M.: MAI. 2011. 278 s.
4. Tarasov E.V., Balyk V.M., Loginov A.B. Vybor opublikovannykh kharakteristik sistem LA. M.: MAI. 1998.
5. Suslov V.I., Ibragimov N.M., Talysheva L.P., Cyplakov A.A. EHkonometrija. Novosibirsk: SO RAN. 2005. 744 s.
6. Afifi A.A., EHjjzen S.P. Statisticheskijj analiz: Pokhod s primenenie EHVM / Per. s angl. M.: Mir. 1982. 488 s.