To reduce the performance uncertainty caused by traffic accidents in a road network, the rnpartial stochastic user equilibrium was used to describe the status of a road network under the impact of rnaccidents, and a bi-level programming model for reliable network design problem was formulated. In rnthe upper level model, the expected total travel time ( ETTT) is adopted as a reliability measure, and rnthe optimization objective is to minimize the weighted sum of the investment costs and the ETTT. The rnlower level model predicts the traffic flow pattern during accidents by generalizing the partial user rnequilibrium model to account for not only the partial impact of traffic accidents but also the imperfect rninformation that travelers perceive. A Monte Carlo simulation-based genetic algorithm was provided to rnsolve the bi-level model. The results of a numerical example show that, compared with the existing rndeterministic method, the proposed model can yield more robust solutions; under three demand levels rnof 100, 150, 200 veh/h, the average increase of social costs is 18％ , but the increase rate of social rncost decrease with the increase of demand.%为减少交通事故导致的路网性能不确定性,基于部分随机用户平衡描述交通事故影响下的路网状态,建立了可靠性网络设计问题的双层规划模型.上层模型将期望总行程时间作为路网可靠性指标,并将其与投资费用的加权和作为优化目标;下层模型综合考虑出行者信息的不完备以及交通事故影响的局部性,扩展了部分用户平衡模型,描述交通事故发生时的交通流模式.用基于蒙特卡洛模拟的遗传算法求解模型.算例表明:与已有确定性网络设计方法相比,本文方案鲁棒性更好,在需求水平为100、150和200辆/h时,社会成本平均增加了18%,社会成本的增加率随交通需求的增加而减少.
展开▼
