基于非支配排序差异演化的应急资源多目标分配算法

摘要

Emergency resource allocation (ERA) is a key topic in emergency management for sudden natural disasters,which mainly deals with how to reasonably and efficiently allocate the emergency relief supplies at reserve points to dispatch points.However,when an extraordinarily serious natural disaster occurs,each dispatch point may ask for many different emergency relief supplies at multiple reserve points at the same time,which will bring potential conflicts over emergency resources.To tackle this problem,a multi-objective optimization model is constructed considering multiple reserve points,multiple dispatch points,multiple emergency resources and emergency resource conflicts resolution.In addition,a multiobjective optimization algorithm for ERA is developed by using none-dominated sorting based differential evolution and encoding repair mechanism.Finally,comparative experimental results from large-scale samples show that our approach can deal with the ERA problem from an overall point of view,simultaneously give the allocation schemes of multiple reserve points for multiple dispatch points,realize different reserve points cooperate with each other on ERA for different dispatch points without any emergency resource conflict,which may provide a useful attempt to solve large-scale ERA problems under limited emergency resources.%应急资源分配(Emergency resource allocation,ERA)是灾害应急管理中的核心环节,主要研究如何高效合理地把各储备点的应急救援物资分配给各发放点.然而,在大规模突发灾害发生后,每个发放点极可能会同时向多个储备点请求多种救援物资,从而带来潜在的应急资源冲突.为此,本文首先构建了考虑应急资源冲突消解的多储备点、多发放点、多种救援物资的应急资源多目标优化模型,并提出了一种基于非支配排序差异演化和编码修正机制的应急资源多目标分配算法.对比实验结果表明,该算法在大规模样本下能够从全局角度同时给出多个发放点的应急资源分配方案,有效实现多个储备点同时为多个发放点协同配备应急资源,而且不会产生任何应急资源冲突,为解决应急资源受限情况下的大规模应急资源分配问题提供了一个有益的尝试.