面向地对空防御的分布式协同任务规划研究OACHSSCD

Facing the challenges of large-scale heterogeneous collaboration and communication constraints in ground-to-air defense systems under strong adversarial environments,this paper proposes a hierarchical distributed cooperative mission planning framework.In the pre-combat stage,a distributed asynchronous multi-round auction(DAMA)algorithm based on a block information sharing strategy is designed.Through a virtual node negotiation mechanism,it breaks through the resource allocation bottleneck under partially connected communication topologies,reducing communication overhead by approximately 38％while ensuring global consistency in allocation.During the combat stage,an improved multi-agent deep deterministic policy gradient(IM-MADDPG)decision model integrating a multi-head attention mechanism and asynchronous experience update is constructed.This model enhances the relational modeling capability of local neighborhood features and effectively alleviates the oscillation problem in policy learning under complex game environments.Simulation results show that the proposed method achieves a static mission completion rate of 96.8％±1.2％,and the dynamic interception success rate and asset survival rate increase to 89.6％±1.5％and 91.2％±1.3％,respectively.It realizes closed-loop optimization from resource pre-positioning to real-time gaming,providing theoretical support for the construction of intelligent defense systems in communication-constrained environments.