【作者】 夏萍；

【导师】 刘凯；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2010， 博士

【摘要】 我国是一个自然灾害频发的国家,地震、飓风、冰冻等重大自然灾害造成巨大的人员伤亡和财产损失,必然需要大量的应急物资解决灾民基本生活、卫生防疫及灾后重建等问题,从而产生了巨大的物流需求。自然灾害发生后,在尽可能短的时间内实现有限物资的最优分配,提出应急物流快速响应过程中的应急物资分配策略,建立适用于救灾工作需要的物资分配决策模型体系,对于实时、快速地制定应急物资动员方案,最大限度地发挥已有资源的使用效益、减轻自然灾害的影响具有重要意义。本文针对灾害应急物流特征,重点研究自然灾害态势下基于需求分析的应急物资分配问题,主要内容包括以下几个方面。(1)灾害应急物流物资管理基本理论研究阐明应急物流的内涵与特征、灾害应急物流的定义和特征；对应急物资管理基本内容进行概括,给出了应急物流物资网络三级结构。从分配特征、影响因素、分配原则三个方面对应急物资的分配问题进行分析。(2)灾害应急物资需求预测研究分析归纳应急物资需求的主要内容及其特征。针对应急物资需求特点,研究应急物资需求预测方法。建立基于多元回归的应急物资需求估测模型和基于支持向量机(SVM)的应急物资需求估测模型,并以汶川地震为例,通过仿真实验验证其有效性,为科学的进行物资需求预测提供技术方法支持。(3)灾害应急物流应急物资需求级别划分针对传统分类管理方法存在的问题,研究应急物资分类管理新方法。通过对影响物资需求分级的指标因素的定性分析,指出物资的重要性、时效性和缺口度是影响物资需求分级的三个关键因素,在此基础上提出把概率神经网络与模糊规则用于应急物资需求级别划分决策中,建立基于概率神经网络的应急物资需求分级模型。并通过算例进行验证,基于概率神经网络的分级方法可以很好的解决应急物资的分级问题,为决策管理人员提供科学分级的理论依据。(4)基于灾区需求与优先级划分的应急物资分配模型构建在对影响灾区需求的关键因素进行分析的基础上,建立灾区需求属性聚类分析模型,进而对灾区群组优先级排序,帮助将应急物资有的放矢、优先分配给需求最急迫的地区。基于目标规划理论,建立基于灾区需求与优先级划分的应急物资分配决策模型,使得整个应急物资的分配过程最优。以汶川地震灾害为例,对物资分配决策进行验证,丰富和完善应急物资调配决策的优化理论和技术。并基于控制反馈原理对应急物资分配动态决策过程进行分析。(5)应急物资分配方案的实施与运作本文从组织结构、信息网络及应急物流快速反应运作模式三个方面对应急物资分配方案的实施与运作的支撑体系进行构建。提出基于供应链管理思想构建跨功能的应急组织结构,分析了应急物资保障信息支撑体系存在的问题并给出合理建议,探讨了应急物流信息系统构成形式,并建立具有高度适应能力的、自我调节的和动态的应急物流快速反应运作模式。对救援人员的指派问题进行了研究,构建了基于遗传算法的模糊多目标灾害指派问题的数学模型,并通过算例验证其可行性。更多还原

【Abstract】 China is a country prone to natural disasters, earthquakes, hurricanes, ice and other natural disasters which caused huge casualties and property losses, and certainly requires a lot of emergency materials for victims to resolve the basic living, health and epidemic prevention and reconstruction issues, resulting in massive logistical needs. To realize the optimal distribution of the limited materials within the shortest time possible after natural disasters happened, to propose distribution decision of emergency materials in the process of emergence logistics quick response, develop an appropriate distribution decision model system of relief goods and materials needed,which have great significance on establishing mobilization scheme of emergency materials promptly and timely, maximizing efficiency in the use of existing resources, reducing the impact of natural disasters is important.In this paper, focusing on the characteristics of natural disaster emergency logistics, the distribution of emergency materials are investigated based on the needs analysis in the situation of natural disasters, The main contents include the following aspects.(1) Research on the basic theory of materials management in natural disaster emergency logisticsThe meaning and characteristics of emergency logistics, definition and characteristics of disaster emergency logistics are clarified; the basic content of emergency materials management is summarized, tertiary structure of the network for emergency logistics materials is given. The issues of emergency materials distribution are analyzed from the three aspects:the characteristics of distribution, influencing factors and distribution principles.(2) Research on the natural disaster emergency materials demand forecastThe main contents and characteristics the needs of emergency materials are summarized on the basis of the characteristics of emergency materials. Focusing on the demand characteristics for emergency materials and the problems of traditional, the demand forecasting methods for emergency materials are studied. The demand forecasting models of emergency materials based on multiple regression and support vector machine-based (SVM) 1 are established. Taken the earthquake happened in Wenchuan as example, the simulation results demonstrate that it is effective and can afford technical support for conducting emergency material demand forecast. (3) Research on the prioritization of emergency materials needs in disaster emergency logisticsFocus on the problems in the traditional classification management, the new classification method of emergency materials is studied. After qualitatively analyzing of the index factors which impact on materials demand classification, the three key elements for materials classification are pointed out:the materials importance, the materials urgency and the materials scarcity, probabilistic neural network (PNN) and fuzzy rules are applied in the decision-making, the model of emergency materials demand classification is built based on the probabilistic neural network. And an example verify the approach based on probabilistic neural network can solve the classification issue for emergency materials, it can provide theoretical basis for the scientific classification.(4) Construct the emergency materials distribution model based on the disaster needs and priorities divisionbased on the analysis In the key factors affecting the demand for disaster areas, the disaster needs attribute clustering model is established, and thus prioritizing disaster groups, to help emergency materials will be targeted, priority assigned to the most urgent needs of the area. Based on the goal programming theory, emergency materials distribution model is set up on the basis of the disaster area needs and priorities division of, making the whole process of the emergency materials distribution optimal. Taken the Wenchuan earthquake as example to verify emergency materials distribution decision, it proved that it can enrich and improve the deployment of emergency materials in decision-making optimization theory and techniques. After that, the dynamic decision-making process of the emergency materials distribution is analyzed based on feedback control theory.(5) Study on the operation and implementation of the emergency materials distribution decision.Three support system on the operation and implementation of the emergency materials distribution decision are constructed, it includes the organizational structure, information network and emergency logistics rapid response operation; cross-functional emergency organization structure is proposed based on supply chain management, of emergency materials support information support system problems are analyzed and reasonable proposals are given, emergency logistics information system constitutes are discussed and a highly adaptable, self-regulation and fast dynamic response of emergency logistics operation is establish. The issue of the rescuer assignment is studied, fuzzy multi-objective mathematical model based on genetic algorithm of disaster assignment problem is constructed, and an example to validate its feasibility.更多还原