V.V Goncharov - Dr.Sc. (Eng.), Professor, Head of Mathematical Department, Peter the Great Strategic Rocket Forces Academy (Moscow). E-mail: v_v_goncharov@mail.ru O.V. Mishenina - Ph.D. (Ped.), Professor, Department of Mathematics, Peter the Great Strategic Rocket Forces Academy (Moscow). E-mail: o.v.misyenina@gmail.com

The purpose of work perfects the system of a long-term forecast; succeed the demand level of competences students on the basis on rational set definition of controllable parameters and optimum duration of untestable intervals of periodic control of knowledge-s mathematical formation process. Considering knowledge of the student as object, and their formation, ? as management process, the problem of an as-sessment of the train quality is reduced to a problem of an assessment of ready difficult technical system to objective per-formance on mission. To each controllable parameter (to the section, a subject) there corresponds a certain feature set as which values of measured signals (level of training) from control points - sensors of controllable object (students) are used. Forecasting is carried out with use of the device of artificial neural networks (ANN). The method of sliding windows is applied to formation of training sample of ANN. The neural network is trained on these samples and forms as result demanded function of a forecast. Process of forecasting occurs similar to process of formation of training sample. For an assessment of accuracy of model it is offered to carry out a retrospective forecast. As a measure of a quality forecast the factor of a divergence is used. For realization of this method of forecasting it is used by multilayer perception. The package of applied programs is necessary for realization of this pro-gnostic model on the Neural Network Toolbox of MATLAB system. The offered model allows to carry out a long-term forecast of formation of students - mathematics, namely, to predict an exit of the most significant controllable competences to demanded level. The obtained data can be used for determination of optimum duration of intervals between periodic tests of a condition of training process in educational institution.

 

1. Identifikacija i tekhnicheskaja diagnostika / Pod red. A.I. Polousa. M.:VA RVSN, 2012. 257 s.
2. Krug P.G. Nejjronnye seti i nejjrokompjutery. M.: Izdatelstvo MEHI. 2002. 176 s.
3. CHetyrkin E.M. Statisticheskie metody prognozirovanija. M.: Statistika. 1977. 200 s.
4. KHajjkin Sajjmon. Nejjronnye seti. Polnyjj kurs. M.: Izdatelskijj dom «Viljams». 2006. 1104 s.
5. Ogandzhanjan S.B., Rozhnov A.V., Burmistrov P.A., Lobanov I.A., Tjurin S.A. Tvorcheskie materialy «kruglogo stola». CHast I. Retrospektiva i realnaja konkordancija issledovanijj v sfere intellekta // Nejjrokompjutery: razrabotka, primenenie. 2016. № 1. S. 17-29.
6. JAzyk skhem radikalov: metody i algoritmy / Intellektualnye informacionnye sistemy // pod red. A.V. CHechkinai A.V. Rozhnova. Kollektivnaja monografija. CH 1. M.: Radiotekhnika. 2008.
7. Rozhnov A.V. Nekotorye problemnye voprosy sistemnojj integracii napravlenijj nauchnojj dejatelnosti v zadachakh nejjrokompjutinga // Nejjrokompjutery: razrabotka, primenenie. 2014. № 1. S. 3-9.
8. Kublik E.I., Rozhnov A.V. Sistemnaja integracija napravlenijj nauchnojj dejatelnosti v uslovijakh formirovanija predyntellektualnojj infrastruktury // Informacionno-izmeritelnye i upravljajushhie sistemy. 2014. № 11. S. 59-64.