Comparative Analysis of Multi-Agent Methods for Constrained Global Optimization

This paper considers the application of three metaheuristic methods of constrained global optimization, which use the ideas of evolutionary algorithms and methods of “swarm” intelligence. Two of them are bioinspired, i.e. algorithms that imitate the behavior of certain animal kinds. Based on the described algorithms, software is developed in Microsoft Visual Studio. The program allows to find a constrained global extreme of functions of two variables with a complex structure of the level lines and visualize the step-by-step process of finding solutions, compare the efficiency of the procedures used. Based on the comparative analysis of algorithms, a hybrid multi-agent method was proposed and its effectiveness is searched.