Experience in Using Stochastic Optimization Methods for Determining Numerical Parameters of Models in Materials Structurization Management Systems

The task of intellectual support of the process of composition formation for materials with a composite structure occurs when designing and operating automated control systems for multi-stage production processes. Such automated systems function in direct interaction with the external environment, and should promptly return the results of processing to the environment in the form of corrective actions or as messages to the user. The need for correct and complete mathematical models and fast, accurate algorithms that solve multistage problems often arises when structuring composite materials. In this case, mathematical models contain sets of numerical parameters and the search for exact values for them presents a complex optimization problem. The purpose of this paper is to investigate the possibility of using stochastic optimization methods to determine the exact numerical values of the calculated parameters of mathematical models that mimic the behavior of a structured composite material with given physico-mechanical characteristics under operating conditions. To carry out the research, special software has been created that implements algorithms for searching for extreme values for functions of several variables. The functional purpose of the software is intellectual support for decision-making in the formation of chemical compositions of cast iron alloys. Another developed system is designed to make effective decisions when designing the composition and structure of composite materials containing discrete fibers. Optimization of the calculated parameters was performed on a definite and fixed search area, which is a hyperparallelepiped. The program implements ten modifications of the simulation algorithm for annealing, allowing for a finite number of steps to make an estimate of the optimal value of the input elements of the function under study on a multidimensional space. In particular, modification of A, B and B algorithm schemes using the Boltzmann and Cauchy distribution functions, as well as the superfast annealing algorithm and the Xin Yao algorithm are implemented. The obtained data allowed to draw conclusions about the advantages and disadvantages of each modification of the stochastic search algorithm..