基于资源自动分配大规模粒子群算法  

Large-scale Particle Swarm Optimization Based on Automatic Resource Allocation

作　　者：邱小品 陈得宝[1] QIU Xiaopin;CHEN Debao(School of Physics and Electronic Information,Huaibei Normal University,235000,Huaibei,Anhui,China)

机构地区：[1]淮北师范大学物理与电子信息学院,安徽淮北235000

出　　处：《淮北师范大学学报(自然科学版)》2025年第1期38-44,共7页Journal of Huaibei Normal University:Natural Sciences

基　　金：国家自然科学基金项目(61976101);安徽省学术和技术带头人后备人选科研活动经费项目(2021H264);安徽省高校学科(专业)拔尖人才学术资助项目(gxbjZD2022021)。

摘　　要：为实现计算资源有效分配,提出一种基于资源自动分配大规模粒子群算法(VIDRA_MMOPSO)。利用随机分组对变量进行分组,得到不同变量平均位置组成的新个体,对新个体同组变量施加一定幅度随机扰动,确定每组变量重要度,根据不同组变量重要度,设计资源自动分配模型,实现不同变量组计算资源自动分配,提升大规模粒子群优化算法性能。9个标准测试函数仿真实验表明,VIDRA_MMOPSO算法在反世代距离和超体积2个性能指标上优于大部分对比算法。In order to realize the effective allocation of computing resources,a large-scale particle swarm algorithm based on automatic resource allocation(VIDRA_MMOPSO)is proposed.The variables were grouped by random grouping to obtain a new individual composed of the average position of different variables,and a certain amplitude of random disturbance was applied to the variables of the same group of new individuals to determine the importance of each group of variables.According to the importance of different groups of the variables,a model of automatic allocation of resources was designed to realize automatic allocation of computational resources of different groups of variables,and the performance of large-scale particle swarm optimization algorithm.The 9 standard test function simulation experiments show that the VIDRA_MMOPSO algorithm outperforms most of the comparative algorithms in inverted generational distance and hyper volume.

关 键 词：大规模粒子群算法 资源自动分配 扰动 变量重要度 多目标优化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]