多事故多救援站点的应急车辆调度问题中，在处置当前事故时，若将空闲车辆再配置于救援站点，有利于对潜在事故的快速响应． 文中采用双层规划理论和非合作博弈理论建立应急车辆调度与再配置模型． 上层模型在事故需求和救援时间窗约束下，最小化当前事故响应时间;下层模型将各救援站点视为非合作博弈的局中人，综合考虑车辆再配置时间和救援站覆盖区域潜在风险，确定局中人的收益函数，将优化再配置策略转化为寻求非合作博弈的纳什均衡． 然后，提出一种层次混合蛙跳算法，其中上层算法用于求解约束单目标规划问题，下层算法用于求解非合作博弈模型． 求解事故算例证明了应急车辆调度与再配置模型的合理性和层次混合蛙跳算法的有效性．

When an emergency vehicle scheduling problem involving multiple accidents and multiple rescue sites occurs,the response time of potential accidents can be shortened by redistributing idle emergency vehicles on res- cue sites,in addition to optimizing the scheduling of vehicles for current accidents.This paper presents an improved model of emergency vehicle scheduling and reallocation on the basis of bi-level programming and non-cooperative game.The upper level of the model is established under the constraints of accident requirements and accident rescue window to minimize the response time for current accidents,while in the lower level of the model,each rescue site is treated as a participant in a non-cooperative game,the payoff function of each participant is determined after tak- ing into account the reallocation time of the vehicle and the potential risks within the coverage area of each rescue site,so that the optimal reallocation strategy is transferred into Nash equilibrium in a non-cooperative game.After- wards,an integrated bi-level shuffled frog-leaping algorithm is proposed,which contains an upper-layer algorithm for single-objective programming and a lower-layer algorithm for solving the non-cooperative game.Several illustra- tive examples verify the rationality of the proposed model and the effectiveness of the integrated bi-level shuffled frog-leaping algorithm.