Information Exchange and Adaptive Control in Ant Colonies for Solving Optimization Problems

The article discusses approaches to improving the ant colony algorithm (ACO) in dynamic and decentralized environments. The main focus is on optimizing the mechanisms of information exchange and adaptive management in the tasks of routing, planning and resource allocation. Three key approaches were identified: local interaction between agents, dynamic parameter tuning, and integration of reinforcement learning based on Q-learning. Each of the approaches was analyzed in terms of efficiency, computational costs, and adaptability to tasks with changing conditions. As part of the experiments conducted on the basis of a five-node network, a small-scale assessment was carried out. The results showed that for networks with moderate dynamics and an average number of nodes, the optimal approach is local interaction, minimizing computational costs and speeding up adaptation. In conditions of high environmental variability, dynamic parameter setting provides a more flexible and adaptive solution. For tasks where it is important to take into account experience and predict future conditions, the Q-learning approach is most effective.