摘要
通过正交试验方法来研究影响遗传算法对巡回旅行商问题的求解能力的因素,通过极差分析法和方差分析法得出了影响因素从强到弱依次为交叉率、群体规模、选择算子、变异率;最优参数组合方案为:群体规模500,选择率1%,交叉率40%,变异率1%;结果表明,遗传算法具有较好的鲁棒性。
Orthogonal tests were designed to study parameters and their combination of traveling salesman problem（TSP）.Based on the range analysis and the variance analysis,it figured out that the influential factors from primary to secondary were crossover rate,population size,selection rate and mutation rate.And the optimal case was as follows,Population Size=500,Selection Operator=2%,Crossover Rate=40%,Mutation Rate=3%.It was also verified that the genetic algorithm has good robustness.
出处
《武汉理工大学学报（信息与管理工程版）》
CAS
2010年第3期386-389,共4页
Journal of Wuhan University of Technology：Information & Management Engineering
关键词
巡回旅行商
正交试验
遗传算法
参数组合
traveling salesman problem
orthogonal experimental design
genetic algorithm
parameters combination
