基于网络节点聚类的多无人机动态目标分配

Dynamic Target Assignment of Multiple Unmanned Aerial Vehicles Based on Clustering of Network Nodes

为解决分布式的多无人机目标分配算法容易产生通信冗余,导致编队出现通信规模过大的问题,提出一种基于通信网络节点聚类的多无人机动态目标分配算法(clustered consensus-based bundle algorithm,CU-CBBA)。该算法引入了通信网络节点分组聚类策略,依据节点的度中心性、特征向量中心性、中介中心性等属性,建立了网络节点重要性排序模型,选取网络拓扑结构中的一组关键节点,按照最短路径原则完成网络拓扑节点聚类。仿真结果表明:与CBBA、ACBBA(asynchronous consensus-based bundle algorithm)、MCDGA(minimizing communications in decentralized greedy task allocation)算法相比,CU-CBBA算法的通信规模最小,收敛速度最快,迭代稳定性强,编队规模或目标数量的改变不会影响其有效性,与其他CBBA变种算法相比,该算法不需要智能体之间建立相对稳定的通信拓扑结构,通用性和稳定性较好,对复杂战场环境下的多机多目标分配具有部分借鉴意义。

In order to solve the problem that the distributed multi-UAV target assignment algorithm is prone to communication redundancy,which leads to the large communication scale of formation,a multi-UAV dynamic target assignment algorithm(CU-CBBA)based on node clustering in communication network is proposed.The algorithm introduces the communication network node grouping clustering strategy.According to the node's degree centrality,feature vector centrality,intermediate centrality and other attributes,the network node importance ranking model is established.A group of key nodes in the network topology structure are selected and the network topology node clustering is completed according to the shortest path principle.The simulation results show that,compared with CBBA,ACBBA and MCDGA algorithms,CU-CBBA algorithm has the smallest communication scale,the fastest convergence speed,and strong iterative stability.The change of formation size or target number does not affect its effectiveness.Compared with the other CBBA variant algorithms,the algorithm does not need the establishment of relatively stable communication topology between agents and has good universality and stability,which has a certain reference significance for the multi-aircraft and multi-target allocation in complex battlefield environment.