Analysis of Model and Simulation for UAVs Equipment Swarm Attack-Defense Tactics Based on Non-Static Bayesian Architecture

The drone swarm attack and defense game has increasingly caught on in the study of complex systems and holds significant application significance in military network security and artificial intelligence fields. Due to the uncertainty of the environment and the diverse behaviors of intelligent agents, this game faces difficulties in modeling in actual simulations and requiring intelligent agents to respond quickly and effectively to changes in the situation. This paper proposes an innovative framework focusing on the UAVs swarm attack and defense confrontation between the red and blue sides. Firstly, based on the modified Lanchester equation game model, the KM algorithm is expanded to adaptively handle the multi-target assignment problem during the adversarial process. Secondly, to enhance the environmental adaptability of UAVs, drones swarm attack and defense tactical strategies are designed and using a non-static Bayesian network architecture for real-time inference and prediction to deal with uncertainty in the confrontation, we reduce model complexity, and improve the accuracy and speed of decision-making. Finally, a game model simulation platform is constructed to vividly demonstrate the confrontation dynamics of the red and blue sides’ UAVs equipment swarm, verifying the effectiveness of the proposed algorithm. The simulation results confirm that the proposed theoretical framework can effectively simulate the confrontation process between the red and blue sides, solving the key problem of multi-task assignment.