Dynamic Ontology for Semantic-prognostic Self-organization of a UAV Swarm

The paper proposes a dynamic ontological model for the semantic-predictive self-organization of an unmanned aerial vehicle (UAV) swarm. A multi-layer ontological model of the UAV swarm is developed, comprising a static core, a dynamic instance layer, an agent layer with local knowledge projections, and a mechanism for semantic synchronization of agents to ensure knowledge consistency in a distributed multi-agent system. A mathematical model is formulated to formalize the processes of knowledge alignment, scenario forecasting, and optimal action selection. The organization of interactions between structural layers is examined, providing knowledge exchange, adaptive role allocation, and support for collective agent behavior. Experimental simulation conducted in the Gazebo environment for forest fire monitoring by a group of 12 UAVs confirms the effectiveness of the proposed approach: semantic consistency reached 91%, the accuracy of critical event prediction increased by 17–20%, and network load decreased by 83.4%. The developed model ensures scalability, robustness to the loss of individual agents, and efficient execution of cooperative tasks, making it suitable for the development of intelligent control systems for autonomous UAV swarms.