用混合元启发算法求解城市网络设计问题

为更有效求解城市道路交通网络设计问题,在启发式算法研究的基础上,使用3种改进思想,改进蚂蚁群算法,设计了4种求解城市道路交通网络设计的双层规划模型的混合启发式改进算法。运用于Sioux Falls网络进行模型的求解实验,并统计运行的平均计算时间,求得最优解的次数和函数解平均值。据此得出混合算法从时间、准确度上均较基本蚂蚁算法有了改善,具有很好的应用价值。

Solving the Discrete Network Design Problem with Implemented Hybrid Heuristic Algorithm

To better solve the urban transportation network design problem, three different intelligence concepts were used to enhance the ant system algorithm based on the existed heuristic algorithms. With the ant system algorithm and these three improvement concepts, four kinds of hybrid-heuristic algorithms were devised for the bi-level programming model. Their capabilities were evaluated on a well-known test network of Sioux Falls. The parameters, such as the average computation time of the statistics, the frequency of finding the optimal solution in 50 runs, and the average value of function, were calculated. It is observed that the method could shorten the time to find the better solutions so that they were more effective in solving the network design problem compared with the basic ant system algorithm.