基于最小最大后悔值的应急救灾网络构建鲁棒优化模型与算法

应急救灾过程分为两个阶段：第一阶段启动应急救灾网络构建，在灾区附近设立临时应急配送中心，并由应急资源供应方向其紧急调配应急资源；第二阶段将应急资源从临时应急配送中心向灾区受灾点进行调度，以保证救灾过程顺利进行。本文研究第一阶段应急救灾网络的构建问题，考虑到突发灾害初期灾情相关参数概率分布情况难以获取，建立了基于情景的最小最大后悔值准则的应急救灾网络构建鲁棒优化模型。求解模型时，利用有限情景集表示第二阶段的不确定性数据，并将模型化为与其等价的混合整数规划模型，利用情景松弛的迭代算法进行求解。数值试验中给出相应的绝对鲁棒模型与本文偏差鲁棒模型作了比较，结果表明基于最小最大后悔值准则的应急救灾网络优化模型具有良好的鲁棒性，而且算法也是有效的。

Emergency Network Model and Algorithm Based on Min-Max Regret Robust Optimization

After emergency, rescue process is represented by two stage. First stage is to start the emergency rescue network, including building temporary distribution centers, allocating reasonable emergency resource to assure rescue process working smoothly. The second stage is to dispatch emergency resource from temporary distribution centers to demand points. A min-max regret robust emergency network model is proposed for natural disasters under uncertainty, taking the second stage as auxiliary decision. Uncertain information is random and represented by finite scenarios. The proposed model is solved using scenario relaxation algorithm. Finally, a case study is proposed to highlight efficiency of the proposed model and solution algorithm.