求解TSP的学习记忆果蝇算法  

Novel Fruit Fly Algorithm Based on Learning and Memory for Solving Traveling Salmesman Problem

作　　者：王勇臻 陈燕[1] 张金松[1] 

机构地区：[1]大连海事大学交通运输管理学院,辽宁大连116026

出　　处：《小型微型计算机系统》2016年第12期2722-2726,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(71271034)资助;国家社科基金项目(15CGL031)资助;辽宁省自然科学基金项目(2014025015)资助

摘　　要：提出一种解决TSP的学习记忆果蝇算法.为克服传统果蝇算法易于陷入局部极值的缺点,该算法引入了果蝇的学习记忆机制,通过设置种群记忆区指导果蝇个体的搜索方向,以提高算法的全局寻优能力.同时,为增强算法求解TSP的局部搜索能力,结合λ-opt算法设计了一种离散型嗅觉搜索过程.实验结果分析表明,本文所提出的算法能够有效地求解TSP,具有可靠的全局收敛性,较快的收敛速度.A novel fruit fly algorithm based on learning and memory for solving traveling salesman problem （ TSP ） is proposed. In or- der to overcome the premature convergence drawback of traditional fruit fly algorithm, a learning and memory mechanism of drosophi- la is introduced. With the aim of improving its global optimization ability,the algorithm sets a population memory area so as to guide the search direction of drosophila individuals, which improves its global optimization ability. For enhancing its local search ability for TSP, the λ-opt algorithm is integrated into the algorithm to design a new discrete osphresis foraging phase. The result of TSP examples shows that the algorithm can solve the TSP effectively and it has an excellent search performance no matter in convergence speed or precision.

关 键 词：果蝇算法 旅行商问题 学习记忆机制 λ-opt算法 离散型嗅觉搜索 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]