Abstract: In order to study the impact of risk on vehicle routing choice behavior in congested traffic network, the mathematical model of conditional value-at-risk shortest path problem in stochastic traffic network was established through defining the conditional value-at-risk as the objective function of path. The subadditivity of conditional value-at-risk of path was proved, and the conditional value-at-risk shortest path problem was transformed into the conditional value-at-risk shortest path problem based on road segments. A labeling algorithm based on dynamic programming was constructed to solve the problem. Based on the numerical experiments of Sioux Falls network, the calculation results of the conditional value-at-risk shortest path in stochastic traffic network under different risk confidence levels were compared and analyzed. Numerical results show that the conditional value-at-risk shortest path obtaining under different risk confidence levels is different, and the level of risk confidence has a significant impact on the choice of the optimal path.

Key words: traffic and transportation engineering, stochastic network, optimal path, conditional value-at-risk, labeling algorithm

摘要： 为了研究风险性对于拥挤交通网络车辆的路径选择行为的影响，定义条件风险值为路径目标函数，建立随机交通网络环境下最小条件风险路径问题数学模型，证明了路径的条件风险值的次可加性，把最小条件风险路径问题转化为基于路段的最小条件风险路径问题，构造基于动态规划的标号算法求解该问题，针对Sioux Falls Network展开数值试验，对在不同风险置信水平条件下随机交通网络最小条件风险路径的计算结果进行了比较分析。结果表明：不同风险置信水平条件下求解的最小条件风险路径是不同的，风险置信水平对最优路径的选择具有重大影响。

关键词: 交通运输工程, 随机网络, 最优路径, 条件风险值, 标号算法