摘要
由于不同领导者位置关系的不确定性,传统的多领导者蜂拥算法会在局部产生不利于系统收敛的虚拟吸引力.针对这个问题,引入一个排斥势能函数,并在此基础上提出一种改进的多领导者蜂拥算法.通过理论分析,证明了算法的稳定性和速度的一致性,并证明各个子系统的位置中心将最终趋于其对应领导者的位置.最后,通过仿真验证了算法的有效性和定理中的结论.仿真统计结果表明,在智能体初始位置随机分布的多智能体系统中,本文提出的算法使得系统的渐近收敛时间更短、更稳定.
Due to the uncertainty of different leaders＇position, traditional flocking with multiple leaders would generate virtual attractive force among agents with different leaders, which is against the global flocking. Considering this, this paper introduces a repulsive potential function and proposes an improved flocking algorithm with multiple leaders. Through theoretical analysis, the stability and velocity matching property are proven. Furthermore, we prove that the center position of each sub-system would converge to their corresponding leader＇s position. At last, the numerical study verifies the effectiveness of the proposed algorithm and correctness of theoretical deduction. The statistical results indicates that algorithm in this paper would make the converging time to be shorter and more stable in multi-agent system with random distributed agents.
出处
《小型微型计算机系统》
CSCD
北大核心
2014年第9期2116-2121,共6页
Journal of Chinese Computer Systems
基金
航空科学基金项目(20110580002)资助
中央高校基础研究基金项目(ZYGX2009J092)资助
关键词
多领导者
蜂拥算法
势能函数
稳定性分析
multiple leader
flocking algorithm
potential function
stability analysis
