摘要
为了克服群集蜘蛛优化(SSO)算法易陷入局部极值和收敛速度慢等缺陷,提出了一种融合差分进化和粒子群优化算法搜索机制的改进群集蜘蛛算法。在群集蜘蛛算法中引入差分变异算子,让部分雌蜘蛛进行由种群中全局最优个体和两个随机个体所引导的变异,从而增加种群多样性,提高算法跳出局部最优解的能力。另外,在上述改进的基础上,借鉴粒子群优化算法的搜索机制,在位置更新公式中添加一组动态的非线性惯性权重及学习因子,以更好地平衡算法的局部和全局搜索能力。实验结果表明:改进的群集蜘蛛算法具有更快的收敛速度和更好的求解精度。
In order to overcome the disadvantages of social spider optimization(SSO)algorithm such as easily falling into local optima and slow convergence speed, an improved SSO algorithm integrated with searching mechanisms of differential evolution(DE)and particle swarm optimization(PSO)is proposed.The differential mutation operator of DE is introduced in the social spider optimization algorithm to allow some female spiders to mutate under the guidance of the global optimal individual and two randomly chosen individuals in the population.This mutation strategy can increase the diversity of the population, improve the ability to jump out of local optima, and accelerate the convergence speed of the algorithm.In addition, based on the above improvements and inspired by the searching mechanism of PSO,a set of dynamic nonlinear inertial weights and learning factors are added to the updating equations to get better balance between the local and global search abilities of the algorithm.The experimental results show that the improved social spider optimization algorithm has faster convergence speed and better solution precision.
作者
向蕾
鲁海燕
胡士娟
沈莞蔷
XIANG Lei;LU Haiyan;HU Shijuan;SHEN Wanqiang(School of Science,Jiangnan University,Wuxi 214122,China;Wuxi Engineering Technology Research Center for Biological Computing,Wuxi 214122,China)
出处
《传感器与微系统》
CSCD
北大核心
2021年第6期121-125,132,共6页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61772013,61402201)
中央高校基本科研业务费专项资金资助项目(114205020513526)。
关键词
群集蜘蛛优化算法
差分变异算子
无约束优化
惯性权重
学习因子
social spider optimization(SSO)algorithm
differential mutation operator
unconstrained optimization
inertia weight
learning factor
