摘要
针对饥饿游戏搜索算法(hunger games search, HGS)存在收敛速度慢和易陷入局部最优等缺点,提出了一种基于均衡池和莱维飞行的饥饿游戏搜索算法(equilibrium Lévy hunger games search, ELHGS)。该算法首先利用tent映射产生更具多样性的初始种群;受到平衡优化器算法(EO)的启发,提出一种基于动态均衡池收敛的更新公式,其动态调整的更新策略使算法的全局搜索能力增强;为了进一步增强算法跳出局部最优的能力,在一定条件下对种群实施基于莱维飞行的变异操作。对23个基准函数进行仿真实验,结果显示与原始HGS算法相比,ELHGS求解精度更高、收敛更为迅速,在高维度多峰函数问题上效果最为显著。
Aiming at the disadvantages of the HGS algorithm such as slow convergence and the tendency to fall into local optima,this paper presented a hunger games search algorithm based on equilibrium pooling and Lévy flight.It employed tent mapping to generate a more diverse initial population.Inspired by the equilibrium optimiser algorithm(EO),this paper proposed an update formulation based on the dynamic equilibrium pool convergence principle,whose dynamically tuned update strategy strengthened the global searching ability.To step out of the partial optimum furthermore,it performed a mutation operation based on Lévy flight on population for certain condition.Simulation experiments were carried out for 23 benchmark functions.The results show that compared with original HGS algorithm,ELHGS has higher accuracy and converges more rapidly,which is obvious for high-dimensional multi-peaked functions.
作者
张大明
赵彦清
徐嘉庆
Zhang Daming;Zhao Yanqing;Xu Jiaqing(Guangxi Key Laboratory of Embedded Technology&Intelligent System,College of Information Science&Engineering,Guilin University of Technology,Guilin Guangxi 541004,China)
出处
《计算机应用研究》
CSCD
北大核心
2023年第5期1368-1374,共7页
Application Research of Computers
基金
国家自然科学基金资助项目(52178468,52268023)
广西自然科学基金联合资助培育项目(2019GXNSFAA245037)
广西青年创新人才科研专项资助项目(桂科AD19245012)
广西“嵌入式技术与智能系统”重点实验室开放基金资助项目(2019-02-08)
桂林理工大学博士启动基金资助项目(GUTQGJJ2019042,GUTQDJJ2019041)。
