Abstract: The subject of the research is the task of group management of autonomous agents in a dynamic multi-agent system and self-study of the management model. The author examines such aspects of the problem as a group interaction, using the example of the most effective group control algorithms, such as SWARM, ant algorithm, bee algorithm, firefly algorithm and fish school movement algorithm, and training of an artificial neural network through the use of reinforcement training. A comparison of various algorithms for finding the optimal path. The comparison was made on the basis of the gaming environment "Battlecode", which dynamically forms a new map for the new round, which ensured the quality of the comparison of the considered algorithms. The author uses statistical methods of data analysis, the selection and analysis of qualitative signs, forecasting methods, modeling method, classification method. The author shows that Q-learning increases its effectiveness by replacing the tabular representation of the Q-function with a neural network. This work proves the effectiveness of the bee algorithm in solving the problem of researching and patrolling the area. At the same time, the path search algorithm A* is much more flexible and efficient than the Dijkstra algorithm.

Abstract: The object of this study is strategic planning in the management process. The authors note the importance of innovation and strategic planning in forming an innovative economy. The article gives detailed review of the main tool of strategic analysis - mathematical modeling. The authors make a conclusion about the need to use a more flexible mathematical apparatus, such as the apparatus of game theory. The authors also note that the theory of games is a very complex area of scientific knowledge. It is especially important to understand its application boundaries. A simplified interpretation of the apparatus of game theory can be very dangerous. The theoretical analysis has been used as a method of research. Individual aspects, features and characteristics of mathematical modeling in strategic planning were identified and reviewed. The main conclusion of this paper is to identify and clearly marked a promising field of research. Since the prospects and hopes placed by the authors on the apparatus of game theory for strategic planning aimed at the transition to an innovation economy, can significantly increase the efficiency of the state management of the economy and to ensure its rapid transition from a resource-production to the economy.