Genetic algorithm for solving the inverse problem of chemical kinetics

Modeling allows to get a description of the object being modeled, in particular complications, which improves and clarifies its understanding and presents an organization of information that makes it easier to get the necessary information. Modeling of chemical kinetics problems is the solution of systems of ordinary nonlinear differential equations with determination of kinetic parameters. The inverse problem refers to incorrectly set tasks and does not have a single solution. The development of modern methods of evolutionary optimization, including the genetic algorithm, makes it possible to obtain a solution to such a high-dimensional problem in an acceptable time.

Materials and methods. The object of research is the catalytic reaction of dimethyl carbonate (DMC) with alcohols in the presence of hexacarbonyl tungsten. The solution of the direct problem is the solution of a system of ordinary nonlinear differential equations with initial data and given kinetic parameters. The solution of the inverse problem consists in determining the kinetic parameters corresponding to the minimum deviation of the calculated values of substance concentrations from the experimental data.

Results. The inverse problem of chemical kinetics is solved, which consists in calculating kinetic parameters with minimizing the functional deviation of the calculated values of component concentrations from experimental data. In the language of Python, program has been developed that implements the genetic algorithm. Numerical experiments with different numbers of iterations are performed and their impact on the accuracy of the solution, as well as the corresponding time costs, is estimated.

Discussion and conclusions. The calculated values of the component concentrations based on the kinetic parameters found correspond to the values of the experimental data.

Numerical experiments with different numbers of iterations are performed. Based on experiments, it was concluded that with an increase in the number of iterations, the accuracy of the solution increases. An increase in the number of iterations, in turn, leads to an increase in the time required for performing calculations. The time dependence on the number of iterations is analyzed.

1. Koledina, K.F. Kinetics and mechanism of catalytic reaction of alcohols with dimethyl carbonate / K.F. Koledina, S.N. Koledin, N.A. Shchadneva, I.M. Gubaidullin // Journal of Physical Chemistry. 2017. Vol. 91. No. 3. pp. 422-428.

2. Koledina, K.F. Automated system for identifying conditions for homogeneous and heterogeneous reactions in multi-purpose optimization problems / K.F. Koledina, S.N. Koledin, I.M. Gubaidullin // Siberian Journal of Computational Mathematics. 2019. Vol. 22. No. 2. pp. 137-151.

3. Koledin, S.N. Information system for assessing the relationship of objective functions and the study of optimal conditions for complex catalytic reaction by multi-purpose optimization methods / S.N. Koledin, K.F. Koledin, A.P. Karpenko, I.M. Gubaidullin // Electrical and information complexes and systems. 2017. Vol. 13. No. 4. pp. 71-80.

4. Genetic algorithms / L. A. Gladkov, V. V. Kureychik, V. M. Kureychik // M.: Fizmatlit. 2006. 87 p.

5. Genetic algorithms for solving extreme problems / D.I. Batishchev // Nizhny Novgorod State University. Nizhniy Novgorod. 1995. 62 p.

6. Automatic programming / N.I.Polikarpova, A. A. Shalyto // M.: Peter. 2010. 167 p.

7. Genetic algorithms, artificial neural networks and problems of virtual reality / G. K. Voronovsky, K. V. Makhotilo, S. N. Petrashev, S. A. Sergeev // Kharkiv, OSNOVA. 1997. 112 p.

8. Tikhonova, M.V. Parallel conditional global optimization in mathematical modeling of the kinetics of chemical reactions / M.V. Tikhonova, V.V. Ryabov, S.I. Spivak, I.M. Gubaidullin // Computational methods and programming: new computing technologies. 2013. Vol. 14. No. 1. pp. 262-268.

9. Koledin, S.N. Optimal control and optimum sensitivity in problems of chemical kinetics / S.N. Koledin, K.F. Koledin // Journal of the Middle Volga Mathematical Society. 2016. Vol. 18. No. 3. pp. 137-144.

10. Gubaidullin, I.M. Sequential-parallel determination of kinetic parameters / Gubaidullin I.M., Koledina K.F., Spivak S.I. // Journal of the Middle Volga Mathematical Society. 2009. Vol. 11. No. 2. pp. 14-25.