对抗条件下的无人集群目标重分配方法研究  

Task Reallocation Method for Unmanned Swarm Under Adversarial Conditions

作　　者：张伦 杨妹 赵拓 张水库 黄健[1] Zhang Lun;Yang Mei;Zhao Tuo;Zhang Shuiku;Huang Jian(College of Intelligence Science and Technology,National University of Defense Technology,Changsha 410073,China;PLA 63867 Troops,Baicheng 137000,China)

机构地区：[1]国防科技大学智能科学学院,湖南长沙410073 [2]中国人民解放军63867部队,吉林白城137000

出　　处：《系统仿真学报》2025年第1期1-12,共12页Journal of System Simulation

摘　　要：异构无人集群在未来战争中具有重要的应用潜力,然而,在高强度的对抗中,集群个体遭受打击后,如何高效快速地将受损智能体携带的任务进行重分配进而保证集群任务顺利完成,是无人集群作战运用中必须面对的难题。提出了一种基于改进合同网-匈牙利算法的目标重分配方法。该方法通过指派机制和招标机制,相比对比算法实现了较低通信代价、较快速度的受损智能体目标重分配。仿真实验中,在单个智能体受损情况下,提出的方法将目标重分配完成度平均值由87%提升到100%,在多个智能体受损时仍能稳定完成目标重分配,且该方法资源调度率和快速性均优于对比算法。Heterogeneous unmanned swarms have important potential applications in future wars.However,during the high-intensity confrontation in the battlefield,how to efficiently and quickly redistribute the tasks carried by the damaged agents so that the swarms could successfully complete the mission is a difficult problem that must be addressed in the combat application of unmanned swarms.This paper proposes a task reallocation method named improved CNP-HA(contract net protocol-Hungarian algorithm).Through the allocation mechanism and the bidding mechanism,the method realizes the task reallocation of damaged agents with lower communication cost and faster speed comparing with baseline methods.In the simulation experiment,when a single agent is damaged,the average completion rate of target reallocation is increased from 87%to 100%by the proposed method,and the target reallocation can still be completed stably when multiple agents are damaged,and the resource scheduling rate and rapidity of the proposed method are better than the comparison algorithms.

关 键 词：对抗条件 个体受损 无人集群 目标分配 重分配 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]