A.F. Albu, Yu.G. Evtushenko, V.I. Zubov. Application of second-order optimization methods for solving an inverse coefficient problem in the three-dimensional statement ... P. 19-34

An inverse problem of finding a temperature-dependent thermal conductivity coefficient of a substance is considered. The analysis is based on the first boundary value problem for the three-dimensional nonstationary heat equation. The sample of the substance under investigation has the form of a rectangular parallelepiped. The inverse coefficient problem is reduced to a variational problem. The root-mean-square deviation of the calculated heat flux on the surface of the body from the experimentally obtained flux is chosen as the cost functional. The paper investigates the possibility of solving the variational problem by optimization methods of the second order of convergence. On the example of a number of nonlinear problems whose coefficients are temperature-dependent, a comparative analysis of the solution of these problems by means of the gradient method and the Levenberg–Marquardt method is carried out. The accuracy of calculating the elements of the Jacobi-type matrix required to implement the Levenberg–Marquardt method has a significant impact on the convergence of the iterative process. It is essential that in our approach the elements of the Jacobi-type matrix are calculated with machine precision due to the use of the fast automatic differentiation technique. Much attention is paid to the features of solving the inverse problem associated with its three-dimensional spatial nature.

Keywords: inverse coefficient problems, nonlinear problems, three-dimensional heat equation, optimal control, numerical optimization methods, alternating direction schemes

Received May 10, 2021

Revised June 17, 2021

Accepted June 28, 2021

Funding Agency: This work was supported by the Russian Science Foundation (project no.  21-71-30005).

Alla Filippovna Albu, Dr. Phys.-Math. Sci., Federal Research Center “Computer Science and Control” of Russian Academy of Sciences, Moscow, 119333 Russia, e-mail: alla.albu@mail.ru

Yury Gavrilovich Evtushenko, Academician RAS, Prof., Federal Research Center “Computer Science and Control” of Russian Academy of Sciences, Moscow, 119333 Russia, e-mail: yuri-evtushenko@yandex.ru

Vladimir Ivanovich Zubov, Dr. Phys.-Math. Sci., Prof., Federal Research Center “Computer Science and Control” of Russian Academy of Sciences, Moscow, 119333 Russia, e-mail: vladimir.zubov@mail.ru

REFERENCES

1.   Kozdoba L.A., Krukovskii P.G. Metody resheniya obratnykh zadach teploperenosa [Methods for solving inverse heat transfer problems]. Kiev: Naukova Dumka Publ., 1982, 358 p.

2.   Alifanov O.M. Obratnye zadachi teploobmena [Inverse heat transfer problems]. Moscow: Mashinostroenie Publ., 1988, 280 p. ISBN: 5-217-00134-8 .

3.   Vabishchevich P.N., Denisenko A.Yu. Numerical methods for solving the inverse coefficient problem. In: Methods of mathematical modeling and computational diagnostics, Moscow: MGU Publ., 1990, pp. 35–45 (in Russian).

4.   Samarskii A.A., Vabishchevich P.N. Difference methods for solving inverse problems of mathematical physics. In: Fundamental principles of mathematical modeling, Moscow: Nauka Publ., 1997, pp. 5–97 (in Russian).

5.   Samarskii A.A., Vabishchevich P.N. Vychislitel’naya teploperedacha [Computing heat transfer]. Мoscow: Editorial URSS Publ., 2003, 784 p. ISBN: 5-354-00234-6 .

6.   Marchuk G.I. Adjoint equations and analysis of complex systems. Netherlands: Springer, 1995, 468 p. doi: 10.1007/978-94-017-0621-6 . Original Russian text published in Marchuk G.I., Sopryazhennye uravneniya i analiz slozhnykh sistem, Moscow: Nauka Publ., 1992, 334 p.

7.   Balazs Czel, Gyula Grof. Inverse identification of temperature-dependent thermal conductivity via genetic algorithm with cost function-based rearrangement of genes. Int. J. Heat Mass Transf., 2012, vol. 55, no. 15, pp. 4254–4263. doi: 10.1016/j.ijheatmasstransfer.2012.03.067 

8.   Miao Cui, Kai Yang, Xiao-liang Xu, Sheng-dong Wang, Xiao-wei Gao. A modified Levenberg–Marquardt algorithm for simultaneous estimation of multi-parameters of boundary heat flux by solving transient nonlinear inverse heat conduction problems. Int. J. Heat Mass Transf., 2016, vol. 97, pp. 908–916. doi: 10.1016/j.ijheatmasstransfer.2016.02.085 .

9.   Matsevityi Yu.M., Alekhina S.V., Borukhov V.T., Zayats G.M., Kostikov A.O. Identification of the thermal conductivity coefficient for quasi-stationary two-dimensional heat conduction equations. J. Eng. Phys. Thermophysics, 2017, vol. 90, no. 6, pp. 1295–1301. doi: 10.1007/s10891-017-1686-7 

10.   Albu A.F., Zubov V.I. Identification of the thermal conductivity coefficient in the three-dimensional case by solving a corresponding optimization problem. Comput. Math. Math. Phys., 2021, vol. 61, no. 9, pp. 1416–1431. doi: 10.1134/S0965542521090037 

11.   Albu A.F., Zubov V.I. Determination of the thermal conductivity from the heat flux on the surface of a three-dimensional body. Comput. Math. Math. Phys., 2021, vol. 61, no. 10, pp. 1567–1581. doi: 10.1134/S096554252110002X 

12.   Evtushenko Y.G. Optimizatsiya i bystroe avtomaticheskoe differentsirovanie [Optimization and fast automatic differentiation]. Moscow: Computing Center of RAS Publ., 2013, 144 p. ISBN: 978-5-91601-082-4 .

Cite this article as: A.F. Albu, Yu.G. Evtushenko, V.I. Zubov. Application of second-order optimization methods for solving an inverse coefficient problem in the three-dimensional statement, Trudy Instituta Matematiki i Mekhaniki UrO RAN, 2021, vol. 27, no. 4, pp. 19–34; Proceedings of the Steklov Institute of Mathematics (Suppl.), 2022, Vol. 317, Suppl. 1, pp. S1–S15.