•  
  •  
 

Chemical Technology, Control and Management

Abstract

Methods and simplified computational schemes for optimizing data processing for systems operating in conditions of limited a priori information, changes in the characteristics of external influences, uncertainty of parameters have been developed. To describe a non-stationary object, non-linear identification models are considered, constraints, input conditions for obtaining possible values ​​of output variables are defined. An approach aimed at using identification technologies based on generalization of capabilities of dynamic models, neural networks (NN), mechanisms for regulating variable computing schemes of structural network components, as well as learning algorithms of the NN is proposed. A generalized algorithm for learning NN based on the synthesis of the mechanisms of probabilistic, heuristic and stochastic search, the formation of a set of training data, regulation of the values ​​of variables of linear and nonlinear dependences of inputs and outputs is developed. The efficiency of the implemented data processing software complex is analyzed by various methods, algorithms, speed, graphs and requirements, and also estimates of the stability of the generalized learning algorithm of NN and the minimum mean-square error are obtained.

First Page

54

Last Page

61

References

1. Kuznecov E.S. Upravlenie tehnicheskimi sistemami: Uchebnoe posobie / MADI (TU) - M.: 2001. -262 s. 2. YAdy'kin I.B., SHuyskiy V.M., Ovsepyan T.A. Adap­tivnoe upravlenie neprery'vny'mi processami. -M.: E`nergoatomizdat, 1985. - 240 s. 3. Miroshnik I.V., Nikiforov V.O., Fradkov A.L. Nelineynoe i adaptivnoe upravlenie slojny'mi dinamicheskimi sistemami. - S-Pb.: Nauka, 2000. -314 s. 4. Karabutov N.N. Adaptivnaya identifikaciya sis­tem. Informacionny'y sintez. -M.: Kom Kniga, 2006. - 384 s. 5. Neyrosetevaya model' analiza tehnologicheskih vremenny'h ryadov v ramkah metodologii Data Mining / A. M. Vul'fin, A. I. Frid // Informacionno-upravlyayusch'ie sistemy'. 2011. № 5. -S. 31-38. 6. CHipiga A.F., Voronkin R.A. Obuchenie iskusstvenny'h neyronny'h setey putem sovmestnogo ispol'zovaniya metodov lokal'noy optimizacii i geneticheskih algoritmov // Izvestiya TRTU. T. 33. №4. -S. 172-174. 7. YArushkina N.G. Osnovy' nechetkih i gibridny'h sistem: Uchebnoe posobie. - M.:Finansy' i statistika. 2004. 320 s. 8. Jumanov I.I., Islomov A.B.Optimizaciya obrabotki izobrajeniy mikroob`ektov na osnove rekurrentnogo obucheniya neyronnoy seti i implikativnogo otbora informativny'h priznakov // «Nauka i mir», Mejdunarodny'y nauchny'y jurnal, Izd-vo «Nauchnoe obozrenie», Volgograd. - №5(33), 2016. - s. 78-81 9. Jumanov I.I., Holmonov S.M. Optimizaciya identifikacii nestacionarny'h ob`ektov na osnove segmentacii vremenny'h ryadov i nastroyki parametrov neyronnoy seti// Jurnal «Vestnik TUIT» - Tashkent, 2016. - №4(40)/2016. - s. 32-41 10. Djumanov O.I. Metody' adaptivnoy obrabotki danny'h na osnove mehanizmov gibridnoy identifikacii s nastroykoy parametrov modeley nestacionarny'h ob`ektov // Jurnal «Problemy' informatiki» SO RAN, Novosibirsk. - № 2(31), 2016. - s. 13-21 11. Djumanov O.I., Kholmonov S.M. The modified model of training of neural networks in computer industrial systems with modules for nonstationary objects images processing // Journal of Korea Multimedia Society, Seoul - Uzbekistan, Tashkent - 2016, № 5. -p.54-58. 12. Holmonov S.M. Adaptivnaya obrabotka danny'h nestacionarnogo processa na osnove modeley nechetkoy logiki i neyronnoy seti // «Himicheskaya tehnologiya. Kontrol' i upravlenie» - TGTU, Tashkent, 2010- № 5. - s. 90-96.

Included in

Engineering Commons

Share

COinS