近年来，地震、海啸、洪水、疫情等突发公共事件频发，造成大量人员伤亡和重大经济损失。灾难发生后，短期内产生的巨大应急物资需求对我国现代应急物资管理水平和响应效率提出了更高要求。同时由于灾害本身的不确定性，使得决策变得更加困难。因此如何搭建高效合理的应急供应网络，使得物资能够迅速、准确地分配到需要的地方，是当前亟待解决的问题。基于此，本文在考虑受灾点优先级的基础上，分别构建了二级和三级供应网络模型，探索了在突发事件背景下的应急物资预定位-紧急采购-分配的问题。根据我国应急物资保障方式的特点，搭建了由出救点和受灾点组成的二级网络和由供应商、出救点和受灾点组成的三级网络。同时在模型中引入优先级概念，通过基于层次分析法和熵权法的分层组合赋权为指标体系设置权重，再利用TOPSIS法为受灾点进行优先级评分，为各个受灾点进行优先级排序。供应网络模型考虑了需求和供给的不确定性、灾害带来的不同程度的道路损毁及多种交通方式，从多出救点、多受灾点、多种物资的视角出发，构建了两阶段、多目标的随机规划模型。模型目标为最小化总成本、最小化总响应时间和最小化缺货率。通过线性加权的方法将三目标问题转化为单目标问题，用小规模数值实验求得了模型的精确解。根据数值实验结果，分析对比了两种模型的优劣。同时通过敏感性分析验证了模型的有效性，并得到了一些管理学启示。为了提高模型在解决大规模实际问题中的应用能力，本研究基于第三代快速非支配排序遗传算法（NSGA-III），设计了针对三级供应网络模型的算法。以2023年河北省暴雨洪涝灾害为例，运用所提算法对大规模问题进行求解，得到了三目标问题的帕累托解集。通过筛选出分别以三个目标为导向的解，为决策者在选择方案时提供基准参考。通过深入分析得出的解，本文为决策者提供了多维度的策略选择基准，并揭示了总成本与缺货率、响应时间之间的权衡关系，从而总结出有效的成本降低路径。
In recent years, the frequent occurrence of emergencies such as earthquakes, tsunamis, floods, and pandemics has led to significant loss of life and substantial economic damage. The immense demand for emergency supplies in the immediate aftermath of disasters has placed higher demands on the modern emergency supply management level and response efficiency of our country. Moreover, the inherent unpredictability of disasters further complicates the decision-making process. Thus, constructing an efficient and rational emergency supply network to ensure the swift and accurate distribution of supplies is an urgent problem to be addressed. Based on this premise, this paper constructs two-tier and three-tier supply network models, considering the priority of disaster-affected areas, to explore the issue of emergency supplies pre-positioning, emergency procurement, and distribution in the context of emergency.In alignment with the characteristics of our country‘s emergency supply security methods, a two-tier network composed of rescue points and disaster-affected areas, and a three-tier network incorporating suppliers, rescue points, and disaster-affected areas were established. Concurrently, the concept of priority was introduced into the model, employing a combined weighting approach based on the Analytic Hierarchy Process (AHP) and the Entropy Weight Method to set weights for the indicator system. Furthermore, the TOPSIS method was utilized to score and rank the priority of disaster-affected areas. The model accounts for the uncertainty of demand and supply, varying degrees of road damage caused by disasters, and multiple modes of transportation, initiating from multiple rescue points, multiple disaster-affected areas, and multiple types of supplies to construct a two-stage, multi-objective stochastic programming model. The model aims to minimize total cost, total response time, and shortage rate. The multi-objective model was simplified into a single-objective task through the application of linear weighting method., obtaining an exact solution through small-scale numerical experiments. Based on the results of numerical experiments, the advantages and disadvantages of the two models were analyzed and compared. Additionally, the model‘s effectiveness was verified through sensitivity analysis, yielding several managerial insights.To enhance the model‘s applicability in solving large-scale practical problems, this study designed an algorithm based on the third generation of fast non-dominated sorting genetic algorithm (NSGA-III) tailored for the three-tier supply network model. Taking the 2023 Hebei Province flash flood disaster as a case study, the proposed algorithm was applied to solve a large-scale problem, yielding a Pareto solution set for the three-objective problem. By filtering solutions driven by the three objectives, a benchmark for decision-makers was provided for selecting strategies. Through in-depth analysis of the derived solutions, this paper offers multi-dimensional strategic benchmarks for decision-makers and reveals the trade-off relationship between total cost, shortage rate, and response time, thereby summarizing effective cost-reduction pathways.
