一种新的多目标粒子群优化算法研究  

ON A NEW ALGORITHM OF MULTI-OBJECTIVE PARTICLE SWARM OPTIMISATION

作　　者：庹军民 

机构地区：[1]上海证券交易所,上海200120

出　　处：《计算机应用与软件》2010年第6期230-232,262,共4页Computer Applications and Software

摘　　要：定义了Pareto最优解及与其相关的一些概念,引入了被广泛应用的改进的单目标PSO(Particle Swarm Optimization)算法。在此基础上提出了MOPSO算法,用改进的Pareto最优解概念排挤不满足约束的解,还采取一种新的寻找全局最优粒子策略。为了方便评估算法的性能,提出三个比较规则:AD、SP和ER。求解三个复杂的测试问题的结果显示,MOPSO能求出数量充足的、分布均匀的Pareto最优解。In this paper we first define the Pareto optimal solution and some concepts correlated to it,and secondly introduce the improved Particle Swarm Optimisation which has been widely applied for solving the single-objective optimisation problems.Based on these,our mult-objective Particle Swarm Optimisation,MOPSO,is proposed,in which the improved Pareto optimal solution concept is used to push aside the solutions not meeting the constraints,and a new strategy to find global optimal particles is adopted as well.Three comparison regulations,AD,SP and ER,are suggested to facilitate the evaluation of the performance of MOPSO.MOPSO is employed to solve three complex testing problems and the results show that MOPSO can obtain enough Pareto optimal solutions with even distribution.

关 键 词：PARETO最优解 粒子群 多目标优化问题 粒子 全局最优粒子 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]