基于相似度的自适应多目标粒子群算法
Adaptive Multi-objective Particle Swarm Optimization Based on Similarity
宋倩 1刘衍民 2张小艳 3张岩松3
作者信息
- 1. 贵州民族大学数据科学与信息工程学院,贵州贵阳 550025
- 2. 遵义师范学院数学学院,贵州遵义 563006
- 3. 贵州大学数学与统计学院,贵州贵阳 550025
- 折叠
摘要
在多目标粒子群优化的设计中,收敛性和多样性的正确管理是获得接近真实帕累托前沿并分布良好的近似解的关键.为了更好地平衡收敛性和多样性,防止算法过早收敛,提出了一种基于相似度的自适应多目标粒子群算法(SAMOPSO).首先,算法利用存档中每个解与其他解之间的相似度和适应度值来维护存档获得高质量的候选解.其次,引入了一种基于各粒子信息的自适应飞行参数调整机制,进一步提高SAMOPSO算法的进化性能.最后,将SAMOPSO算法与六个经典多目标优化算法在15个基准测试函数上进行比较,结果证实了该算法的有效性.
Abstract
In the design of multi-objective particle swarm optimization,the correct management of convergence and diversity is the key to obtain a well-distributed approximate solution close to the real Pareto Front.In order to better balance convergence and diversity and prevent premature convergence of algorithms,an adaptive multi-objective particle swarm optimization based on similarity(SAMOPSO)is proposed.Firstly,the algorithm uses the similarity and fitness values between each solution and other solutions in the archive to main-tain the archive and obtain high-quality candidate solutions.Secondly,an adaptive flight parameter adjustment mechanism based on par-ticle information is introduced to further improve the evolution performance of SAMOPSO algorithm.Finally,the SAMOPSO algorithm is compared with six classical multi-objective optimization algorithms on 15 benchmark functions,and the results confirm the effecti-veness of the algorithm.
关键词
多目标优化/相似度/自适应/收敛性/多样性Key words
multi-objective optimization/similarity/self-adaptation/convergence/diversity引用本文复制引用
基金项目
贵州省进化人工智能重点实验室项目([2022]059)
贵州省数字经济重点人才计划项目(2022001)
出版年
2024
NETL
NSTL
万方数据