救灾物资的需求推演及调度研究

发布时间：2018-09-14 09:37

【摘要】：近年自然灾害频发,如何提高救援效率成为应急物流中的重要问题。本文以此为背景,主要研究了救灾物资的分类及分级、救灾物资需求的推演以及救灾物资的协同调配机制。救灾物资分类及分级研究基于权变理论的相似性,将组织行为学中的费德勒权变模型迁移到灾前考虑经济成本的情况下应急物资的定性分类中,指导灾前救灾物资的准备。考虑到自然灾害发生时救援活动及救灾物资的弱经济性,建立了用灰色白化权函数聚类分析的评估体系,以高效地解决应急物资分级问题。进一步基于某一地区所有可能发生的灾种对应急物资的重要程度进行了分级。针对大型地震发生后伤亡人数和救灾物资需求的预测及推演问题,建立“三步法”推演模型：用灰色生成序列改进BP神经网络,建立预测总伤亡人数灰色BP神经网络；应用Mistcherlich模型动态推演每天的伤亡人数；在此基础上,构建基于时间和库存管理的救灾物资需求实时推演模型。针对BP神经网络的改进和救灾物资实时推演模型进行了重点阐述。模型基于每日伤亡人口的动态预测和前线的伤亡数据更新,对每个供给间隔期运送的物资量进行实时推演。基于救灾物资需求的推演,以最快响应、最少损失和最低成本为目标函数,建立了基于确定物资需求的多种物资、多出救点、多受灾点的复杂应急物资调度模型。提出一种改进的调配机制,即同一救援范围内的出救点之间可以互相调配物资的机制,并分别选择出救区域和受灾区域的枢纽中心,建立轴辐式网络,不仅缓解了运输压力,而且由于枢纽中心的存在,保证了救灾物资供需可控,保证救援活动有序进行。
[Abstract]:In recent years, natural disasters occur frequently, how to improve rescue efficiency becomes an important problem in emergency logistics. Based on this background, this paper mainly studies the classification and classification of disaster relief materials, the deduction of disaster relief material demand and the coordination mechanism of disaster relief materials. Based on the similarity of contingency theory, Federer's contingency model in organizational behavior is transferred to the qualitative classification of emergency materials considering the economic cost before the disaster, which can guide the preparation of disaster relief materials. Considering the weak economy of relief activities and relief materials during natural disasters, an evaluation system based on grey whitening weight function cluster analysis is established to efficiently solve the problem of emergency material classification. Further classification is made based on the importance of all possible disasters in an area to emergency supplies. In view of the prediction and deduction of casualties and the demand for disaster relief materials after a large-scale earthquake, a "three-step" model is established: the grey BP neural network is established to predict the total number of casualties by improving the BP neural network with grey generating sequence; Based on the Mistcherlich model, the real-time model of disaster relief material demand is built based on time and inventory management. The improvement of BP neural network and the real-time deduction model of disaster relief materials are discussed in detail. Based on the dynamic prediction of the daily casualty population and the update of the casualty data in the front line, the model is used to estimate the quantity of goods transported in each period of the supply period in real time. Based on the derivation of the demand for disaster relief materials and taking the fastest response, minimum loss and minimum cost as the objective function, a complex emergency material scheduling model based on the determination of the material demand, multiple rescue points and multiple disaster points is established. In this paper, an improved deployment mechanism is put forward, that is, the mechanism of distributing materials among the rescue points in the same rescue area, and selecting the hub center of the rescue area and the affected area, respectively, and establishing the axis-radial network, which not only alleviates the transport pressure, Because of the existence of the hub center, the supply and demand of relief materials are controlled and the rescue activities are carried out in an orderly manner.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：D632.5;D63

【参考文献】