融合能量周期性递减与牛顿局部增强的改进HHO算法  

Improved harris hawks optimization coupling energy cycle decline mechanism and Newton local enhancement strategy

作　　者：赵世杰[1,2] 高雷阜[1,2] 于冬梅[1,2] 徒君[1,2] ZHAO Shi-jie;GAO Lei-fu;YU Dong-mei;TU Jun(Institute of Optimization and Decision,Liaoning Technical University,Fuxin 123000,China;Institute for Optimization and Decision Analytics,Liaoning Technical University,Fuxin 123000,China)

机构地区：[1]辽宁工程技术大学优化与决策研究所,辽宁阜新123000 [2]辽宁工程技术大学运筹与优化研究院,辽宁阜新123000

出　　处：《控制与决策》2021年第3期629-636,共8页Control and Decision

基　　金：辽宁省科技厅博士科研启动基金项目(2019-BS-118,20170520075);国家自然科学基金项目(51704140);辽宁省教育厅基金项目(LJ2019JL017,LJ2017QL031);辽宁省自然科学基金指导计划项目(2019-ZD-0032)。

摘　　要：为增强栗翅鹰优化算法的全局探索能力和局部开采性能,提出一种融合能量周期性递减机制与牛顿局部增强策略的改进栗翅鹰优化算法(improved harris hawks optimization, IHHO).该算法在传统HHO算法基础上,启发于自然界中鹰与猎物间的多轮围捕-逃逸现象且猎物能量整体上呈现递减态势,进而设计一种猎物能量的周期性递减调控因子并嵌入能量函数中,该机制有利于实现IHHO算法全局探索与局部搜索间的多轮动态迭代平衡.牛顿局部增强策略借鉴牛顿迭代思想构造一种猎物邻域(当前最优解)的局部再搜索,并依概率实现IHHO算法的局部寻优性能改善.数值实验验证了不同能量周期数和局部搜索次数对HHO算法性能的差异性影响、优越的并行迭代寻优性能以及高收敛精度、高维情形(100D~10 000D)的较好适用性.To strengthen global exploration and local exploitation capacity of harris hawks optimization(HHO), an improved HHO(IHHO) algorithm is proposed by coupling the energy cycle decline mechanism and the Newton local enhancement strategy. On the basis of the canonical HHO algorithm, a control coefficient of the prey energy cycle decline mechanism is designed and absorbed into the original prey energy function, which inspires multi-round besiege-escaping phenomena between hanks and prey in nature. On the whole, the prey energy is declining with iterations. This mechanism contributes to dynamically balancing the global and local searching ability of HHO. Meanwhile, in consideration of better property of Newton iteration thought, a kind of Newton local reinforcement strategy is constructed for re-exploiting the local neighbourhood of prey(current optima), which results in improving the local searching performance of the IHHO algorithm with probability. Experimental results show the performance difference influence of different number of cycles and local searching on HHO, the superior parallel iterative optimization ability and convergence accuracy of the proposed algorithm, and its better applicability on high dimension cases(100 D-10 000 D).

关 键 词：智能优化算法 栗翅鹰优化算法 能量周期性递减机制 牛顿局部增强策略 高维优化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]