基于滚动时间窗序列的地震救援药品动态配置决策研究

As is known to all that the drug distribution issue for earthquake response is more complicated and arrives with a stronger sense of urgency in an earthquake response, compared with the problems caused in other types of disasters. It is still a big challenge to make an effective and efficient distribution plan, because of the time urgency, incomplete information, and surging demand. In this project, a planning method, based on the sliding time window series, is studied to solve the drug distribution problems in the earthquake response. In the first place, the features of drug distribution planning are discussed based on research accomplishment. Then this paper presents a method which is designed to generate the series of sliding time window with the consideration of time-varying demands. Secondly, two methods are proposed to determine the effectiveness and robustness of the drug distribution plan, according to the evaluation standards of the drug distribution plan; besides, a coupling mechanism of these two indexes is also discussed. Thirdly, a dynamic planning model, aiming to balance the effectiveness and robustness to their perfection, is established with the consideration of sliding time window series, group information updates, global supply networks and multi-phase sharing mechanism. After this, two algorithms are designed to solve the model. Finally, a simulation study is carried out to testify the models and algorithms. As a result, specific plans of drug distribution will be provided for the emergency planners when an earthquake unfortunately happens, and it will also give some better management insights into the construction of emergency decision system.

相对于一般灾害的应急资源配置问题，地震救援药品配置有其特殊性和更强的时间敏感性。如何在时间紧迫、信息不完备和需求激增等条件下，进行科学决策，仍是应急管理领域的一大挑战。本项目在分析现有研究的基础上，首先探讨地震救援药品配置决策特征，提出震情演化过程中反映救援药品需求时变特征的滚动时间窗序列生成方法；再根据配置策略的适配性和鲁棒性指标，研究供需适配效用和协同鲁棒条件的量化界定方法及其耦合机理；进而应用贝叶斯决策理论等，以滚动时间窗序列为基本约束，以组群信息刷新为技术手段，以配置策略“适配度”和“鲁棒度”的均衡最大为目标，综合考虑时间-空间交叉网络，构建地震救援药品动态配置模型，设计相应算法；并以汶川地震为背景，开展实例仿真分析，从而提出一套面向震情动态演化、紧扣时变需求和多阶段协同适配的救援药品动态配置决策方法，为地震救援实践提供科学的配置方案，并对我国应急决策体系构建提供理论依据。

相对于一般灾害的应急资源配置问题，地震救援药品配置有其特殊性和更强的时间敏感性。如何在时间紧迫、信息不完备和需求激增等条件下，进行科学决策，仍是应急管理领域的一大挑战。本项目在分析现有研究的基础上，首先探讨地震救援药品配置决策特征，提出震情演化过程中反映救援药品需求时变特征的滚动时间窗序列生成方法；再根据配置策略的适配性和鲁棒性指标，研究供需适配效用和协同鲁棒条件的量化界定方法及其耦合机理；进而应用贝叶斯决策理论等，以滚动时间窗序列为基本约束，以组群信息刷新为技术手段，以配置策略“适配度”和“鲁棒度”的均衡最大为目标，综合考虑时间-空间交叉网络，构建地震救援药品动态配置模型，设计相应算法；并以汶川地震为背景，开展实例仿真分析，从而提出一套面向震情动态演化、紧扣时变需求和多阶段协同适配的救援药品动态配置决策方法，为地震救援实践提供科学的配置方案，并对我国应急决策体系构建提供理论依据。研究结果表明：（1）滚动时间窗序列是地震灾害响应中的重要特征，使得应急救援药品配置决策具有动态性和MCRC（多时间窗协同鲁棒）特征，决策者们必须考虑整个时间窗序列前后时间窗口的供需；（2）配置原则极大影响救援药品配置方案，决策者们必须在公平和有效性之间做出明智的选择；（3）应急决策的信息是不完整的，可采用贝叶斯组群信息更新方法融合历史信息和样本信息开展决策；（4）地震灾害响应下的应急救援药品配置是一个全球事件，必须考虑从供应中心到转运中心以及从转运中心到受灾点的两阶段配置过程；（5）在地震救援药品配置决策中，决策时间和决策风险的关系是非常重要的，决策者要在决策失误风险和决策不及时风险之间实现均衡，从而实现配置效应的最大化。