A.E. Misnik - Post-graduate Student, Dept of Computer Engineering, Branch of National Research University «Moscow Power Engineering Institute» in Smolensk. E-mail: anton@misnik.by V.V. Borisov - Dr.Sc. (Eng.), Professor, Dept of Computer Engineering, Branch of National Research University «Moscow Power Engineering Institute» in Smolensk. E-mail: vadim.v.borisov@mail.ru

The combined neuronetwork method for modeling of complex technical systems is considered. This method combines possibilities of analytical and neuronetwork approaches to construction of set of logical, parametrical and compositional neuronet models of system. The article describes the main steps of the proposed method: select elements of a complex technical system; the de-signing of a logical model of the system; the designing of a parametric model of the system; the designing of a composite neuronetwork model of the system; the formation of the neural network model-supervisor. The proposed method is based on a two-level compositional neuronetwork model: the neural network-supervisor is used for monitor the structural errors of model on the upper level; the parametric setting up models of elements of the system is car-ried out on the lower level. The example of compositional neuronetwork modelling of hydraulic system with using of the offered method is described. The method allows to raise accuracy of modelling and this method allows to select of control actions according to the chosen target criterion of operative and qualitative management of complex technical system. The method allows to improve, first, accuracy of modelling, secondly, to choose the decisions according to the chosen target criterion of effective management of complex technical system in the presence of restrictions. The management of the complex technical systems (the based on results of compositional neuronetwork modeling), al-lows to carry out the proved choice of decisions in real or pseudo-real time for the account of modification of logical and pa-rametrical models of system, and also mapping of these changes on model of system.

 

1. Meshalkin V.P., Belozerskijj A.JU., Dli M.I. Metodika postroenija kompleksnojj matematicheskojj modeli upravlenija riskami predprijatija metallurgicheskojj promyshlennosti // Prikladnaja informatika. 2011. № 3 (33). S. 100-120.
2. Komartsova L.G., Strelchenko S.S., Lavrenkov Y.N. Neural-Network Model for Determining the Equilibrium Compositions of Multicomponent Systems// Optical Memory & Neural Networks (Information Optics). 2010. V. 19. № 2. P. 149-153.
3. Misnik A.E. Nejjrosetevoe modelirovanie sistem teplosnabzhenija // Nejjrokompjutery: razrabotka, primenenie. 2012. № 2. S. 62-73.
4. Borisov V.V., Misnik A.E. Kombinirovannyjj nejjrosetevojj sposob modelirovanija dlja operativnogo upravlenija slozhnymi sistemami // Informacionnye tekhnologii. 2012. № 7. S. 69-72.
5. Karpov JU. Imitacionnoe modelirovanie sistem. Vvedenie v modelirovanie s AnyLogic 5. SPb.: BKHV-Peterburg. 2005. 400 s.
6. Orlov A.I. Teorija prinjatija reshenijj. M.: EHkzamen. 2005. 656 s.