自适应并行遗传算法求解非对称性旅行商问题  

Self-adaptive Parallel Genetic Algorithm for the Solution to Asymmetric Traveling Salesman Problem

作　　者：张强[1] 杨成永[1] 姚宝珍[1] 张磊 

机构地区：[1]北京交通大学土木建筑工程学院,北京100044 [2]大连市供水有限公司,辽宁大连116021

出　　处：《物流技术》2010年第12期82-84,共3页Logistics Technology

摘　　要：遗传算法是一种模拟生物进化的搜索最优解算法。根据交叉概率与变异概率在不同时期的不同要求,提出一个随进化代数和群体的适应度进行自适应调整的策略。同时,采用粗粒度并行策略,使算法具有更好的收敛性、稳定性、更快的速度和更好的优化质量。最后通过一些精典的非对称性旅行商问题对该算法进行了检验,结果表明自适应并行遗传算法可以有效提高遗传算法的求解性能。Genetic algorithm is a searching method for the optimal solution that simulates biological evolution.In accordance with the different requirements of crossover probability and mutation probability in different time periods,the paper proposes an algorithm self-adjusting with evolution iteration and population adaptability.Meanwhile,coarse-grain parallel strategy is incorporated so that the algorithm is improved in convergence,stability,processing rate,and optimization quality.Finally the effectiveness of the algorithm is validated through several empirical studies on the canonical asymmetric traveling salesman problem.

关 键 词：遗传算法 粗粒度 自适应策略 

分 类 号：O242.23[理学—计算数学]