动态不确定性环境下大规模抢险救灾应急资源保障网络优化设计方法研究

Designing an emergency support network and making corresponding adjustment in uncertain and dynamic environments are necessary considering realistic requirements in organizing and implementing large scale military operations for disaster relief. However, the current academic research on designing and adjusting such a network is relatively insufficient. This project focuses on the research of optimization methods to design emergency resource support network basing on the various disaster relief tasks undertook by the military. The modeling methods and decision-making flow adapting the dynamic and uncertain environments will be firstly proposed synthetically using uncertain optimization and dynamic decision-making theories. Then, the optimization models, which are utilized to design and adjust the emergency resource support network at the relief mobilization stage, will be developed to identify the initial mobilization and allocation plans of various emergency resources that originate from the military and its subsequent adjustment schemes. Subsequently, the optimization models, which are utilized to design and adjust the emergency resource support network at the relief implementation stage, will also be proposed to identify the locations of various facilities (i.e., temporary depots, field hospitals and helicopter landing sites), the allocation plans of different types of emergency resources and their succeeding adjustment plans. In addition, the heuristic algorithms basing on various strategies, such as problem decomposition and neighborhood search, will also be proposed to solve these optimization models, ensuring that high quality solutions can be obtained quickly and efficiently to satisfy the emergency requirement in making decisions. The academic results of this project can synthetically bring the military key engines in programming and designing the emergency resource support network of post-disaster relief. Furthermore, they can be taken as valuable accumulation on models and algorithms to help the military improve its ability efficiently in responding to traditional and non-traditional security threat challenges.

在大规模抢险救灾军事行动中基于动态不确定性的环境条件开展应急资源保障网络的优化设计及动态调整，是抢险救灾指挥决策中的现实需求，而当前的研究比较薄弱。本项目基于我军在这类行动中所承担的多样化救援任务及应急资源保障需求，开展相应的保障网络优化设计方法研究：综合运用不确定性优化和动态决策等理论，提出适应动态不确定性环境的建模方法和决策流程；建立面向救灾动员的应急资源保障网络优化设计及调整模型，确定武装力量的动用-分配方案及调整计划；建立面向救援实施的应急资源保障网络优化设计及调整模型，确定临时配送中心、野战医院和直升机着陆点等的定位布局-资源分配方案及调整计划；基于问题模型的结构特点，采用问题分解、邻域搜索等策略，提出快速高效的模型求解算法。本项目的研究成果能够为我军系统地开展应急资源保障网络的规划设计提供核心引擎支持，可为提高我军应对传统及非传统安全威胁的能力提供必要的理论依据和方法储备。

为应对突发性重大自然灾害，最大程度减少财产损失，保护人民生命安全，须要在灾后第一时间进行科学高效的应急资源动员和调度。然而，这类活动的组织和实施过程中往往面临多种动态性和不确定性的挑战。本项目面向这些挑战，主要开展了如下三方面的工作：. （1）基于随机规划的灾后应急资源保障网络优化问题研究。针对灾后应急设施选址、救援物资/人员/运输工具的动员、救援资源的分配等问题，考虑灾后救援需求、运输时间、供应水平等多种因素的不确定性，构建了一系列基于情景-概率的混合整数随机规划模型，设计了基于情景分解的启发式模型求解算法。. （2）基于鲁棒优化的灾后应急救援资源动员-分配-运输问题研究。针对灾后物资供应与医疗处置能力动员、物资前送与伤员后撤等需求，分别构建了基于单周期和多周期的应急资源动员-分配-运输优化模型，考虑到实际应用中情景、概率比较难以确定，采用两种典型的鲁棒优化方法，推导得到了相关模型的鲁棒对应形式，开展了鲁棒优化参数敏感性分析，对两种鲁棒优化方法的效果进行了对比；. （3）面向动态不确定性的应急资源运输调度及调整方法研究。基于灾后应急运输所面临的动态不确定性及对运输计划进行实时调整的需求，综合考虑应急物资的运送和伤病员的撤退，建立了基于多周期规划的应急运输调度模型，创新性地引入了基于鲁棒模型预测控制（RMPC）的多周期滚动优化框架，以获取鲁棒的应急运输计划，并能够支持根据实时揭晓的信息对所生成的计划进行调整。基于汶川地震救援案例验证了所提模型和框架的有效性，分析了鲁棒优化参数设置对优化结果的影响。. 本项目的研究成果对于灾后应急救援指挥决策能起到很好的理论和方法支持。