SOLUTION OF THE INVERSE PROBLEM OF CHEMICAL KINETICS BASED ON FUZZY LOGIC MODELS

The possibility of using “soft” calculation methods to solve the inverse problem of chemical kinetics to identify fuzzy kinetic laws based on the data of stationary experiments without a priori specifying the type of the kinetic law has been investigated. “Soft” computing combines artificial intelligence methods such as fuzzy logic, neural networks, genetic algorithms, etc. (F. Rosenblatt, L. Zadeh, J. Holland, etc.). Combinations of these methods will allow the creation of various hybrid systems. Dynamic kinetic models of chemical reactions are described by systems of ordinary differential equations (clear deterministic models), on the basis of which, using various optimization methods (least squares, etc.), the inverse problem of chemical kinetics is solved. B. Kosko proved that any mathematical system can be approximated as accurately as desired by a non-deterministic or weakly deterministic model based on fuzzy logic using simple algorithmic constructions of the “if ... then” type (conditional operators) without direct quantitative or qualitative analysis of systems of differential equations. A fuzzy kinetic model is an algorithmic design based on fuzzy algebra, which allows one to calculate the reaction rate for any values of reagent concentrations of interest to the experimenter. This model allows solving the inverse problem without a priori specifying the type of kinetic law, does not have an unambiguous formulaic representation and can be described by regression (clear) equations. In this article, based on fuzzy logic methods, regression kinetic models are obtained and studied to solve the inverse problem for a simple model reaction and the acetylene hydrochlorination reaction. It has been shown that clear regression models demonstrate good accuracy and can be used to establish alternative kinetic laws and mechanisms for chemical reactions. For citation: Kol'tsov N.I. Solution of the inverse problem of chemical kinetics based on fuzzy logic models. ChemChemTech [Izv. Vyssh. Uchebn. Zaved. Khim. Khim. Tekhnol.]. 2024. V. 67. N 12. P. 54-63. DOI: 10.6060/ivkkt.20246712.7067.