Abstract:

When monadic analytic and multivariate optimisation analysis are used for selecting location of transfer hub in clustered cities, they may have some defects that the results may center on individual cluster. This model consideres the interaction and feedback between transfer hub layout and transport network construction, transfer hub layout and the network traffic flow. The minimum total cost of road network and the maximum relation intensity between clusters are taken as the upper- level model. The user equilibrium traffic assignment model is taken as the lower- level model. A genetic algorithm to solve the model is also designed. The results of an example shows that when uses in a clustered city composed of two clusters and twenty nodes with the condition of at most only one transfer hub being located in each cluster, the GA algorithm needed twenty seconds less than the SAB algorithm. The computational results validate the correctness in the problem description and effectiveness in the solution time of the proposed model.

Key words: traffic engineering, bi- level programming, genetic algorithms, clustered cities, transfer hub, semi fixed location

摘要：

针对一元解析法和多元最优化方法用于组团城市换乘枢纽选址时，结果可能集中个别组团的问题，以路网总成本最小、组团间联系强度最大为上层模型，以用户均衡交通量分配模型为下层模型，构建了组团城市换乘枢纽的双层规划选址模型，设计了求解该模型的遗传算法.算例表明，用于4 个组团、20 个节点的组团城市换乘枢纽选址时，在每个组团最多只布设1 个换乘枢纽的约束下，遗传算法比SAB算法收敛速度快50 s，稳定性高9%，验证了该模型在描述问题时的正确性及求解问题的优越性.

关键词: 交通工程, 双层规划, 遗传算法, 组团城市, 换乘枢纽, 半定型选址