Open AccessTHE EFFICIENCY ANALYSIS OF MULTI-AGENT OPTIMIZATION METHODS OF AIRCRAFT DESIGNS ELEMENTS
Author(s) -
А. В. Пантелеев,
М. С. Каранэ
Publication year - 2019
Publication title -
naučnyj vestnik mgtu ga
Language(s) - English
Resource type - Journals
eISSN - 2542-0119
pISSN - 2079-0619
DOI - 10.26467/2079-0619-2019-22-2-96-108
Subject(s) - metaheuristic , computer science , set (abstract data type) , mathematical optimization , software , penalty method , function (biology) , industrial engineering , algorithm , engineering , mathematics , evolutionary biology , biology , programming language
The article considers the use of three multi-agent methods for optimizing structural elements of aircraft. The research describes strategies for finding solutions to multi-agent metaheuristic algorithms, such as: fish school search, krill herd, and imperialist competition algorithm. The work of these methods is based on the processes occurring in an environment that features many agents. Agents have the opportunity to exchange information in order to find a solution to the problem. These methods allow you to find an approximate solution, but, nevertheless, with great success are used in practice. In this regard, the described metaheuristic algorithms were applied to the optimization problems of structural elements of aircraft such as: welded beam, high pressure vessel, gearbox and tension spring. The article adduces the formulation of these problems: the objective function, a set of constraints and a set of admissible solutions are indicated, recommendations on the choice of parameters of the methods used are given. To solve the problems of optimizing the elements of aircraft construction, a set of software elements was formed in the development environment of Microsoft Visual Studio in C #. This complex of programs allows you to solve the given problems by each of the described multi-agent methods. The software allows you to select a method, a task and select the method parameters and the penalty function coefficients in the best possible way. The results of the solution were compared with each other and with the well- known solution. According to the numerical results of solving these tasks, we can conclude that the algorithmic and software created allow us to find a solution close to the exact one in a reasonable time.