路网最优费率的双层规划模型及算法

Bi-level programming model and algorithm of optimal toll rate for highway network

为寻求系统、科学的路网最优费率测算方法,提出用双层规划模型描述路网管理者、收费道路经营者和车辆用户三者在费率决策中的博弈关系,其中上层以路网用户盈余最大化为目标,下层是弹性需求下的多车型随机用户均衡模型,设计了遗传-模拟退火混合优化的求解算法。计算发现路网收费收入目标的高低将直接影响测算所得费率的高低,进而影响各车型OD出行量,且时间价值低的车型用户比时间价值高的车型用户受费率变化的影响要更为显著,表明双层模型在费率决策过程中能兼顾相关各方的利益,建模时考虑车型分类能更合理地反映现实情况;与遗传算法、模拟退火算法相比,遗传-模拟退火算法计算结果最优。

In order to find system and science method to calculate the optimal toll rate of highway network, a hi-level programming model to determine the optimal toll rate was put forward, the relationship among highway network managers, toll road operators and users was described. Its upper objective function was consumer surplus which should be maximized, its lower-level problem was multi-vehicle-type stochastic user equilibrium model with elastic demand. A kind of hybrid optimization algorithm combined genetic algorithm and simulated annealing to solve it was proposed. Calculation result shows that the value of revenue for highway network influences the toll rate directly, furthermore, it influences OD traffic flows, and the influence on the vehicle types with low time value is more obvious than on the vehicle types with high time value, which indicates that the model can balance the benefits among managers, operators and users reasonably, and reflect the fact more accurately when considering vehicle types; compared with genetic algorithm and simulated annealing algorithm, the computation result of the algorithm for the model is least, the algorithm is feasible. 5 tabs, 1 fig, 12 refs.