大规模公共卫生事件下城市即时配送网络优化模型与算法

Optimization Model and Algorithm of Urban Real-Time Distribution Network in Large-Scale Public Health Emergencies

摘要:

大规模公共卫生事件下城市即时配送存在路网数据失真、供需侧信息不确定及网络中断问题，亟需考虑信息不确定性及路网中断可能性进行城市即时配送网络优化。首先，考虑封控导致道路限行下的路网构建问题，建立城市底层路网并提出改进的Floyd算法；其次，针对开放式多配送点的城市即时配送问题，考虑供需不确定性及设施服务中断问题，使用蒙特卡洛模拟方法构造情景树，建立多目标随机规划模型并设计混合进化算法求解；最后，以2022年上海新冠肺炎疫情事件为例，发现大规模公共卫生事件导致配送设施服务能力、路网容量及客户需求突变，配送系统容易因供需不匹配而发生“爆单”“爆仓”，但一方面设施服务中断未必导致配送成本增加，而是通过降低客户满意度来增加总成本，另一方面更多的车辆使用数目未必导致总成本增加。

关键词:大规模公共卫生事件;中断;即时配送;多目标随机规划模型;蒙特卡洛模拟;混合进化算法

Abstract:

Urban real-time delivery faces challenges such as distorted road network data， uncertain supply-demand information， and network disruptions in large-scale public health emergencies. Therefore， it is essential to consider information uncertainty and network disruptions when optimizing urban real-time delivery networks. To address these challenges， first， the problem of road network construction under road restrictions caused by lockdown measures was addressed， the underlying urban road network was established and the Floyd algorithm was improved for efficient solutions. Then， for open multi-delivery point urban real-time delivery problems， considering supply-demand uncertainty and facility service disruptions， the Monte Carlo simulation was conducted to construct a scenario tree， a multi-objective stochastic programming model was constructed， and a hybrid evolutionary based on GA-SA was designed to solve the problem. Finally， taking the COVID-19 pandemic in Shanghai in 2022 as an example， a comprehensive analysis was conducted. The results reveal that large-scale public health emergencies can lead to sudden changes in delivery facility capacity， road network capacity， and customer demands. The mismatch between supply and demand in the delivery system can result in issues like excessive orders and stockouts. Interestingly， facility disruptions do not necessarily lead to an increase in delivery costs. Instead， the total cost is amplified by a reduction in customer satisfaction. Furthermore， increasing the number of vehicles does not necessarily lead to an increase in costs.

Key words:large-scale public health emergency;disruption;immediate delivery;multi-objective stochastic programming model;Monte Carlo simulation;hybrid evolutionary algorithm