CAJ | 학술논문

目前粒子群优化算法和分布估计算法较少用于解决排列编码组合优化问题，本文提出了一种新的适用于求解排列问题的分布估计离散粒子群优化算法。提出的算法结合粒子群优化算法和分布估计算法的思想，突破了标准粒子群优化算法速度-位移更新模式。新算法中每个粒子的信息一部分来自该粒子当前解排列与全局最优排列的最长公共子串，另一部分来自描述所有个体最优值分布信息的概率模型。这样粒子的当前解、所有个体最优值和全局最优值都参与了新解的生成过程，提出的算法秉承了粒子群优化算法的思想，同时具有更全面的学习能力，提高了算法的寻优能力以及避免陷入局部最优的能力。在两个经典的排列问题上的实验结果表明提出的算法具有良好的性能。
목전입자군우화산법화분포고계산법교소용우해결배렬편마조합우화문제，본문제출료일충신적괄용우구해배렬문제적분포고계리산입자군우화산법。제출적산법결합입자군우화산법화분포고계산법적사상，돌파료표준입자군우화산법속도-위이경신모식。신산법중매개입자적신식일부분래자해입자당전해배렬여전국최우배렬적최장공공자천，령일부분래자묘술소유개체최우치분포신식적개솔모형。저양입자적당전해、소유개체최우치화전국최우치도삼여료신해적생성과정，제출적산법병승료입자군우화산법적사상，동시구유경전면적학습능력，제고료산법적심우능력이급피면함입국부최우적능력。재량개경전적배렬문제상적실험결과표명제출적산법구유량호적성능。
Particle swarm optimization algorithm (PSO) and estimation of distribution algorithm (EDA) are seldom applied to permutation-based combinatorial optimization problems .This paper presents an estimation of distribution-discrete particle swarm optimization algorithm (ED-DPSO ) for the permutation-based problems .In ED-DPSO ,one part of components of the offspring comes from the longest common subsequence between the current solution and the global best solution ,and the other part comes from the probability model built on the distribution information of all personal best solutions .In ED-DPSO ,the current solution ,all personal best solutions and global best solution contribute to the generation of a new solution .Thus ,ED-PSO has more comprehen-sive learning ability ,and can avoid falling into local minima and improve the search ability .Experiment results on two classic per-mutation-based problems show ED-PSO has superior performance .