Abstract: The solution to traffic line selection problem was converted to minimum spanning tree（MST）problem.Based on the graphic theory，an improved genetic algorithm is introduced to search the minimum spanning trees.This algorithm uses binary code to represent the problem of minimum spanning trees and uses the depth first searching method to determine the connectivity of the graph.The corresponding fitness function，single parent transposition operator，single parent reverse operators and controlling evolutionary strategies are designed to improve its speed and efficiency.In comparison with traditional algorithms，it can acquire a set of minimum spanning trees during one genetic evolutionary process.

Key words: traffic line selection, Minimum Spanning Tree（MST）, genetic algorithms, optimization

摘要： 将交通选线问题求解转化为最小生成树（Minimun Spanning Tree，MST）的求解，对比了经典MST求解算法，以图论为基础，采取一种求最小生成树的改进遗传算法．该算法以二进制编码表示最小树问题，用深度优先搜索算法进行图的连通性判断，并采用相应的适应度函数、单亲换位算子和单亲逆转算子及多种控制进化策略，能在一次遗传进化过程中获得一批最小生成树，可供决策部门综合评价与决策．

关键词: 交通选线, 最小生成树, 遗传算法, 优化