中文摘要：

基于城市道路网结构与交通流特征,以总配送耗时最小为目标函数,以交通流为约束条件,构建了双层配送网络优化模型。上层模型计算配送车辆的配送路径,下层模型为用户均衡交通分配模型,通过上层模型的计算结果改变下层模型中的OD出行数据,通过下层模型的计算结果改变上层模型中的路段通行时间。利用混合式分组法、遗传算法与Frank-Wolf算法求解模型,并以大连市某带有31个交通小区、27个需求点和4个配送中心的交通网络为例进行实例验证。计算结果表明：当利用最短距离法求得配送方案时,27个需求点的总配送距离为94.8km,总配送耗时为425.2min,计算时间为13s;考虑交通流变化后,利用提出的双层优化模型,27个需求点的总配送距离为109.7km,总配送耗时为329.1min,计算时间为256s。利用提出的双层优化模型,虽然总配送距离增加14.9km,但总配送耗时却缩短96.1min,并可以一次性达到配送车辆和其他车辆相互平衡的过程,计算速度和效率并不是最重要的因素,可以得到更符合实际的计算结果。

英文摘要：

Based on the structure of urban road network and the characteristic of traffic flow, the minimum total distribution time was taken as objective function, the traffic flow was taken as constraint condition, and the optimization model of bi-level distribution network was developed. The distribution routing of distribution vehicle was calculated by using the upper model, the lower model was user equilibrium traffic assignment model, the OD trip data in the lower model could be changed by using the calculation result of upper model, and the transit time of road section in the upper model could be changed by using the calculation result of lower model. The hybrid grouping, genetic algorithm and Frank-Wolf algorithm were used to solve the bi-level optimization model, and example verification was carried out through a traffic network with 31 traffic zones, 27 demand points and 4 distribution centers in Dalian City. Calculation result shows that when the distribution plan is obtained through the minimum distance method, the total distribution distance for 27 demand points is 94.8 km, the total distribution time is 425.2 min, and the calculation time is 13 s. Using the bi-level optimization model with the consideration oftraffic flow change, the total distribution distance for 27 demand points is 109.7 km, the total distribution time is 329.1 min, and the calculation time is 256 s. After using the bi-level optimization model, the total distribution distance increases 14.9 km, but the total distribution time decreases 96.1 min, the mutual balance process of distribution vehicles and other vehicles can one-off reaches, the calculation speed and efficiency are not the most important factors, and the calculation result more in line with the actual situation can be gotten. 1 tab, 10 figs, 22 refs.