This article is cited in 7 papers

Optimization, System Analysis, and Operations Research

Control of set of system parameter values by the ant colony method

I. N. Sinitsyn^ab, Yu. P. Titov<sup>ba

a</sup> Moscow Aviation Institute (National Research University)
^b Federal Research Center "Computer Science and Control" of Russian Academy of Sciences, Moscow

Abstract: The paper considers the modification and application of the ant colony method for the problem of directed enumeration of the values of system parameters when performing calculated multiple calculations. Interaction with the user makes it possible to stop the process of exhaustive enumeration of sets of parameter values, and the application of a modification of the ant colony method will allow us to consider rational sets at early iterations. If the user does not terminate the algorithm, then the proposed modifications allow one to enumerate all solutions using the ant colony method. To modify the ant colony method, a new probabilistic formula and various algorithms of the ant colony method are proposed, allowing for each agent to find a new set of parameter values. The optimal algorithm, according to the research results, is the use of repeated endless cyclic search for a new solution. This modification allows you to consider all solutions, and at the same time, find all the optimal solutions among the first 5% of the considered solutions.

Keywords: ant colony method, parametric graph, reordering, computing cluster, hyperparameter optimization.

Presented by the member of Editorial Board: F. T. Aleskerov

Received: 23.01.2023
Revised: 21.03.2023
Accepted: 09.06.2023

DOI: 10.31857/S000523102308010X

 English version: Automation and Remote Control, 2023, 84:8, 893–903