首页|基于强化竞争学习鸽群优化的多无人机博弈决策

基于强化竞争学习鸽群优化的多无人机博弈决策

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
  • 维普
多无人机(multi-unmanned aerial vehicle,multi-UAV)博弈决策是无人机博弈对抗领域的关键性技术难题.本文提出了一种强化竞争学习鸽群优化(enhanced competitive learning pigeon-inspired optimization,ECPIO)算法,以解决多无人机博弈决策问题.建立了六自由度无人机模型以及无人机博弈态势评估模型,据此计算博弈双方无人机的收益矩阵.采用非合作博弈模型,将多无人机博弈决策问题建模为基于混合策略纳什均衡的目标函数寻优问题,采用强化竞争学习鸽群优化求取目标函数的最优解,通过引入强化竞争学习策略可有效降低优化结果陷入局部最优的概率.最后,通过将所提出的强化竞争学习鸽群优化算法与基本鸽群优化、基本粒子群优化、竞争粒子群优化以及反向学习粒子群优化算法的仿真对比实验,验证了所提方法解决多无人机博弈决策问题的可行性与有效性.
Decision-making of multi-UAV combat game via enhanced competitive learning pigeon-inspired optimization
Decision-making of multi-unmanned aerial vehicle(multi-UAV)combat game is a crucial problem in the field of unmanned aerial vehicle combat game.In this study,an enhanced competitive learning pigeon-inspired optimization(ECPIO)algorithm is proposed to handle decision-making of multi-UAV combat game.Firstly,a six degree of freedom UAV model is adopted and situation assessment between UAVs is designed,the payment matrixes corresponding to two players in combat game are calculated.Then,non-cooperative game model is selected,the problem of multi-UAV cambat game decision-making is transformed into optimization based on the mixed Nash equilibrium,and ECPIO is adopted to calculate the optimal solution.ECPIO preserves the obvious advantage of pigeon-inspired optimization(PIO),which has fast convergence rate.Our proposed ECPIO can reduce the probability of optimization results trapping into local optimum by introducing enhanced competitive learning strategy.Finally,ECPIO is compared with the basic PIO,basic particle swarm optimization(PSO),competitive particle swarm optimization(CSO)and opposition-based learning particle swarm optimization(OBPSO)by a series of comparative simulation experiments,and the experimental results verify the feasibility and superiority in solving the decision-making of multi-UAV combat game problem.

multi-unmanned aerial vehicle(multi-UAV)combat game decision-makingpigeon-inspired optimization(PIO)enhanced competitive learning

雷阳琦、段海滨

展开 >

北京航空航天大学自动化科学与电气工程学院,仿生自主飞行系统研究组,北京 100083

鹏城实验室,深圳 518000

多无人机 博弈决策 鸽群优化 强化竞争学习

科技创新2030-"新一代人工智能"重大项目国家自然科学基金项目国家自然科学基金项目国家自然科学基金项目国家自然科学基金项目

2018AAA0100803U20B2071T212100391948204U1913602

2024

10.1360/SST-2022-0032

中国科学(技术科学)
中国科学院

中国科学(技术科学)

CSTPCD北大核心
影响因子：0.752
ISSN：1674-7259
年,卷(期)：2024.54(1)
  • 8