Vestnik of Samara State Technical University. Technical Sciences SeriesVestnik of Samara State Technical University. Technical Sciences Series1991-85422712-8938Samara State Technical University61003Original ArticleParametric identification in inverse heat conduction problems under interval uncertainty based on neural networksDiligenskayaA. N.<p>(Dr. Sci. (Techn.)), Professor</p>info@eco-vector.comSamokishA. V.<p>Postgraduate Student</p>info@eco-vector.comSamara State Technical University151220202846181502202115022021Copyright © 2020, Samara State Technical University2020<p>The application of parametric identification methods used to solve inverse problems of technological thermophysics under conditions of interval uncertainty of parameters is considered. To solve such problems under the action of disturbing factors, artificial neural networks can be used that reproduce the structure of the operator of the direct problem with a small dimension of the desired vector of parameters. A variant of constructing a neural network for solving the boundary inverse heat conduction problem with a known structure of a mathematical operator, characterized by a small number of parameters, is proposed. The method allows the use of a priori information about the admissible range of belonging of the identified characteristics or their parameters. The input data for the neural network model are all possible realizations of temperature states, obtained as a reaction to the input action, which is an identifiable characteristic that satisfies all permissible options for the combination of parameters.</p>
<p>The solution of the problem is an iterative procedure, at each step of which, using a neural network, the value of one of the parameters specifying the parametric representation of the desired value is refined. The combination of neural network technologies with high-quality information about the regularities of non-stationary heat conduction processes makes it possible to increase the accuracy of solving the problem of parametric identification under conditions of disturbances. Variants of solving the boundary inverse problem of heat conduction with one- and two-parameters representation of the desired characteristic under the conditions of disturbances are demonstrated. The accuracy of solving the problem is investigated depending on the intensity of the disturbing effect</p>inverse heat conduction problemparametric identificationinterval uncertaintyneural network modelобратная задача теплопроводностипараметрическая идентификацияинтервальная неопределенностьнейросетевая модель[Alifanov O.M. Obratnye zadachi teploobmena [Inverse Heat Transfer Problems]. Moscow: Mashi-nostroenie, 1988, 280 p. (In Russian).][Beck J.V., Blackwell B., C.R. Jr St. Clair. Inverse Heat Conduction: Ill-Posed Problems, New York: Wiley, 1985.][Alifanov O.M., Artyukhin E.A., Rumyantsev S.V. Jekstremal'nye metody reshenija nekorrektnyh zadach [Extremum Methods of Solution of Ill-Posed Problems and Their Application to Inverse Heat-Transfer Problems]. Moscow: Nauka, 1988, 288 p. (In Russian).][Diligenskaya A.N., Rapoport E.Ya. Analytical methods of parametric optimization in inverse heat-conduction problems with internal heat release // J. Eng. Phys. Thermophys., 2014. Vol. 87, no. 5, p. 1126.][Diligenskaya A.N. Methods of Sequential Parametric Optimization in Inverse Problems of Technologi-cal Thermophysics // 2019 XXI International Conference Complex Systems: Control and Modeling Problems (CSCMP). – Samara, Russia, 2019. pp. 267–270.][Pranghishvili I.V., Lototsky V.A., Ghinsberg K.S., Smolyaninov V.V. System identification and control problems: on the way to modern system methodologies // Control Sciences. 2004. № 4. Pp. 2–15 (In Russian).][Voronova N.P., KovalevS.M., Shabelnikov A.N. Identification and estimation of states for fuzzy dy-namical systems // Izvestiya SFedU. Engineering Sciences. 2016. № 6 (179). Pp. 128-130 (In Rus-sian).][Berdnik V.V., Mukhamedyarov R.D. Application of the Method of Neural Networks to Solution of the Inverse Problem of Heat Transfer // High Temperature, 2003, V. 41, pp. 839–843.][Kruglov V.V., Dli M.I., Golunov R.Y. Nechetkaja logika i iskusstvennye nejronnye seti [Fuzzy logic and artificial neural networks]. Moscow: Fizmatlit, 2001. 221 p. (In Russian).][Haykin S. Nejronnye seti: polnyj kurs. 2-e izd [Neural networks: full course. 2nd ed.], Rev. M.: I.D. Williams, 2006. 1104 p. (In Russian).][Aksenov S.V., Novoseltsev V.B. Organizacija i ispol'zovanie nejronnyh setej (metody i tehnologii) [Or-ganization and use of neural networks (methods and technologies)]. Tomsk, 2006 (In Russian).][Lusenko D.S., Danilushkin I.A. Dynamic model of waste heat boiler based on recurrent neural network // Vestn. Samar. Gos. Tehn. Un-ta. Tehn. Nauki, 2020. V. 28, No. 2. Pp. 59–72 (In Russian).][Zamyatin N.V., Medyantsev D.V. Methods of neural network modeling of complex systems // Izvestia Tomsk. Pol-teh Un-ta, 2006. No. 8. Pp. 100–106. (In Russian).][Gorbachenko V.I., Lazovskaya T.V., Tarkhov D.A., Vasilyev A.N., Zhukov M.V. Neural Network Technique in Some Inverse Problems of Mathematical Physics. In: Cheng L., Liu Q., Ronzhin A. (eds) Advances in Neural Networks – Lecture Notes in Computer Science, 2016. V. 9719.][Pavlov D.Yu. Neural network solution of the coefficient inverse heat conduction problem // Vestn. Mosk. Univ., Ser. 15: Vych. Mat. Kibern., 1994, no. 4, pp. 51–56.][Eremenko Y.I., Poleshchenko D.A., Tsygankov Y.A. Development of neural network model of the mul-tiparametric technological object // J. Fundam. Appl. Sci., 2017. 9 (7S). Рр. 706–721.]