基于搜索空间自适应分割的多目标粒子群优化算法  

Multi-objective Particle Swarm Optimization Algorithm Based on Self-adaption Partition of Search Space

作　　者：孙冲 李文辉[1] SUN Chong;LI Wenhui(College of Computer Science and Technology,Jilin University,Changchun 130012,China)

机构地区：[1]吉林大学计算机科学与技术学院,长春130012

出　　处：《吉林大学学报（理学版）》2019年第2期345-351,共7页Journal of Jilin University:Science Edition

基　　金：吉林省科技发展计划项目(批准号:20170204020GX)

摘　　要：提出一种基于搜索空间自适应分割的多目标粒子群优化算法,根据粒子的搜索能力和规模与子搜索空间的体积呈多维标准正态分布变换,精细分割搜索空间,向划分出的子搜索空间分布粒子实现优化,分割在迭代时持续进行,直至获得最优解集.实验结果表明:该方法解决了多目标粒子群优化算法易陷入局部极值的问题;在反向世代距离性能指标上,该算法与一些典型的多目标粒子群优化算法相比,其种群多样性和解的收敛性优势显著.We proposed a multi-objective particle swarm optimization algorithm based on self-adaption partition of searching space,according to the search capability and scale of particles,it was transformed into a multi-dimensional standard normal distribution with the volume of subsearch space,which finely partitioned search spaces and optimized the distribution of particles in the divided subsearch spaces,and the partition continued during iteration until an optimal solution set was obtained.The experimental results show that the algorithm effectively solves the problem that multi-objective particle swarm optimization algorithm is easy to fall into local extremum,compared with some typical multi-objective particle swarm optimization algorithms,the algorithm has significant advantages in diversity of population and convergence of solution on the performance index of inverted generational distance.

关 键 词：粒子群优化 多目标优化问题 多维标准正态分布 自适应分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]