摘要
反向学习策略可以提高自然计算方法性能,然而现有策略生成的反向解多样性不足,为此提出一种反向学习策略,算法根据群体最优个体计算其正交补空间的反向解,增加种群多样性以提高找到全局最优解的概率.将所提策略应用到标准粒子群与标准遗传算法并在基准测试函数上进行验证,实验结果表明策略能提高算法在多数测试函数上的性能.最后,将策略与重心反向学习结合应用于随机拓扑粒子群算法,CEC13函数集作为测试函数,与四种经典或性能优异的反向学习粒子群算法进行对比,实验结果验证了策略的有效性.
Opposition-based learning can improve the performance of natural computing methods.However, the diversity of opposition-based solutions generated by existing strategies is insufficient.Therefore, an opposition-based learning strategy is proposed, the algorithm calculates the opposition-based solution in orthogonal complementary space based on the optimal particles of the group, increasing the diversity of the population to improve the probability of finding the global solution.Apply the proposed strategy to the standard particle swarm and standard genetic algorithm, verify it on the benchmark function.The experimental results show that the strategy can improve the performance of the algorithm on most benchmark functions.Finally, the combination of strategy and centroid opposition-based learning is applied to the random topology particle swarm algorithm, the CEC13 is used as the test function, compare with four classical or excellent particle swarm algorithms that combine opposition-based learning.The experimental results verify the effectiveness of the strategy.
作者
黄亮
张军
季伟东
HUANG Liang;ZHANG Jun;JI Wei-dong(School of Computer Science and Information Engineering,Harbin Normal University,Harbin 150025,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2023年第3期544-552,共9页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(31971015)资助
2021年度黑龙江省自然科学基金项目(JJ2021LH0990)资助
哈尔滨市科技局科技创新人才研究专项项目(2017RAQXJ050)资助
哈尔滨师范大学计算机科学与信息工程学院科研项目(JKYKYY202001)资助。
关键词
正交补空间
反向学习
多样性
粒子群算法
自然计算
orthogonal complementary space
opposition-based learning
diversity
particle swarm optimization
natural computing
