Application of swarm intelligence principles for decentralized management of intelligent agents

The article deals with the question of constructing intelligent real-time agents as systems of bounded rationality. Such systems must have flexible mechanisms for managing their mental resources, allowing them to adapt to the dynamically changing time of decision-making and to use it as effectively as possible to think about solutions. The algorithms of swarm intelligence are considered as a special class of optimization algorithms for decentralized control of the behavior of intelligent agents. The principal differences between the algorithms of swarm intelligence from other stochastic optimization algorithms are indicated, and UML diagrams showing a unified approach to the implementation of algorithms of swarm intelligence are presented.