【作者】 陈达强；

【导师】 刘南；

【作者基本信息】 浙江大学 ， 管理科学与工程， 2010， 博士

【摘要】 我国作为世界上受自然灾害影响最为严重的国家之一,长期深受各类自然灾害的侵扰,各种损失让人触目惊心,为了降低灾害损失、恢复灾区正常社会经济秩序和保障民众正常生活,各级政府和地区进行了各项应急系统建设工作,尤其是应急物流系统及应急物资储备与分配管理。从已有的研究成果分析来看,对由于受灾系统的时间动态变化、出救点运力限制、应急限制期约束以及系统应急响应成本限制等应急系统物资供求特性的物资优化分配问题尚未进行深入研究,反映现有的理论研究与应急管理中的实际需求存在较大差距。事实上,本研究所重点强调的物资时变供求特性——即在灾情变化的不同阶段(主要指随时间变化)应急物资表现出不同的供求特点,是实际应急物流系统的真实情况。因此,从应急物资供求特性分析入手来研究其优化分配问题,其研究更具现实意义。在相关研究成果分析的基础上,本研究主要研究面向大规模自然灾害应急物流运作的核心：应急物资分配模型的构建与求解。通过归纳应急物资分配系统的特性,分析其决策目标与约束条件,构建出符合实际应急运作的优化决策模型,分析模型可行解和最优解条件,并设计求解优化算法。本论文的研究内容主要包括：(1)总结了应急物流系统的内涵和系统结构,分析了应急物资分配系统的系统特性,提出了基于应急物资供求特性分析进行其优化分配的研究思路：以“应急响应时间最早”、“参与出救点最少”为优化目标,将科学问题分解为非时变供求约束与带时变供求约束两个层面,涵盖系统响应成本修正、应急限制期与运力限制三大约束,涉及简单网络与复杂网络两个维度。(2)针对非时变供求约束情形,研究了存在成本修正和物资调运运力限制约束的多出救点选择问题。分析了应急响应的成本组成并将应急响应成本作为方案优劣评价的修正,构建了基于成本修正的应急物流物资响应决策模型,并设计了求解优化算法；针对带运力限制的多出救点选择决策问题,定义了从应急响应指令下达初始至最后一批物资送达需求点的时间为应急响应时间,分析了出救点物资可供应量、供应量和重复运输次数之间的关系,构建了带运力限制约束的多出救点选择多目标决策模型,并设计了相应的优化算法。(3)针对带时变供给约束情形,研究了的多出救点选择问题。创新地提出了“临界应急响应时间”概念,分析了实际临界应急响应时间与理论临界应急响应时间之间的关系,分析了临界应急响应时间与最大运输时间的关系,构建了带时变供给约束的多出救点选择模型,设计了符合带时变供给约束的多出救点选择多目标决策模型的优化算法。在此基础上,研究了非时变供求约束情形下带运力限制约束多出救点选择的时变供给转换模型,并设计了相应的优化算法。(4)针对带时变供求约束情形,研究了带时变供求约束的多出救点选择与单出救点、多需求点应急物资分配问题。分析了带时变供求约束情形下应急系统最小应急响应时间的构成,以及实际临界应急响应时间与理论临界应急响应时间之间的关系,构建了带时变供求约束的多出救点选择多目标决策模型,并设计了求解优化算法。在此基础上,分析了带时变供求约束情形下,单出救点、多需求点应急物资分配方案的构成,创新地提出了“拓扑转换”方法将原问题(单出救点、多需求点)变换为多出救点选择问题,并设计了相应的优化算法。(5)考虑时变供求与应急限制期约束,研究了多需求点应急物资分配模型。分析了带时变供求与应急限制期约束情形下应急方案的构成,分析了系统应急响应时间与各需求点应急限制期之间的关系,创新地提出了在不考虑应急限制期的可行方案基础上对方案进行调整的解决思路,并设计了相应的优化算法。(6)在复杂网络环境下,研究了带时变供求约束的复杂网络(多出救点、多需求点)应急物资分配问题。分析了带时变供求约束复杂网络情形下应急方案的构成,分析了系统应急响应时间与需求点应急响应时间、出救点应急响应时间之间的关系,分析了临界应急响应时间与系统最小应急响应时间之间的关系,提出了约定分配规则,并设计了相应的优化算法。通过对以上问题的研究,可以帮助应急物流运作管理决策者更好地理解应急物流与应急物资分配的内涵,为保障应急物资供应和应急物流运作提供智力支持和决策参考,实现应急物资分配决策的目标,包括应急时间最小和参与出救点个数最少,从而最大限度地降低大规模自然灾害等突发事件对社会正常秩序地冲击,减少人员伤亡和财产损失。更多还原

【Abstract】 As one of the most severe affected area, China suffered great losses from all kinds of natural disasters for a long period. With the objective of reduce disaster losses, recovering the social and economic order and ensuring the normal living condition of victims, central government and local authorities have made a great effort in disaster relief operation and management, especially the emergency logistics and emergency material reserves and allocation management. Most of the previous researches are all carried out without taking dynamic change in time (i.e. emergency material time-varying character), transportation capacity constrain, response time constraint and emergency response cost into consideration, results in the gap between theory research and actual operation requirement. While in reality, emergency material has a time-varying character in supply-demand, i.e. supply-demand character of emergency material will change in different relief operation phase, which is the main topic in this paper. So, starting with the supply-demand character analysis in emergency material allocation research has more sense of realistic value.Based on related research achievements, this paper focused on the core of emergency logistics operation in large-scale natural disasters, i.e. the material allocation model constructing and solving. With the summaries of the supply-demand characters, and the analysis of the decision objectives and constraints, optimal decision models were established, and computation algorithms were designed by feasible solution and optimum solution condition analysis. The main topics in this paper are as following:(1) With summaries in the connotation and systemic structure of emergency logistics and analysis of emergency material allocation’s systemic characters, a basic idea of systematic research in this paper was proposed. In this idea, taking the earliest emergency response start time and the minimum number of relief material distribution center involved (short as R2DC) as bi-objective, the problem was decomposed into time-varying supply-demand constrained problem and non-time-varying supply-demand constrained problem in simple network and complex network structure, with emergency response cost constrain, response time constraint and transportation capacity constraint.(2) Non-time-varying supply-demand constrained problem was studied with two sub problems:emergency material allocation model with emergency response cost constrain and material allocation model with transportation capacity constrain. Firstly, by analyzing emergency response cost elements and considering emergency response cost as a modification as the assessment of the solutions, an optimal decision model for material allocation in emergency logistics with emergency response cost constrain was established and computation algorithm was designed. Secondly, the emergency response start time was defined as the time that the last relief material transported to the affected area and relief operation activity starts. With the analysis of the relationship among material supply amount, the available amount and the number of transportation repeated, material allocation model with transportation capacity constrain and optimized combinatorial algorithm were designed.(3) Multi-R2DC selection problem with time-varying supply constraint was studied. Firstly, the concept of critical response time was proposed. Based on the analysis of difference between theoretical critical response time and actual critical response time, and the relationship between critical response time and max transportation time, a multi-objective decision model for multi-R2DC selection with time-varying supply was established with an optimized combinatorial algorithm. Secondly, the methodology for multi-R2DC selection problem with time-varying supply constraint was applied into the non-time-varying supply-demand with transportation capacity constrain problem, and a new optimized combinatorial algorithm was designed(4) Time-varying supply-demand constrained problem was studied with two sub problems:multi-R2DC selection problem and single-R2DC and multi-affected-area emergency material allocation problem. Firstly, by analyzing earliest emergency response start time elements, and the difference between theoretical critical response time and actual critical response time, an optimal decision model for multi-R2DC selection problem with time-varying supply-demand constraint was established and computation algorithm was designed. Secondly, the methodology for multi-R2DC selection problem with time-varying supply-demand constraint was applied into single-R2DC and multi-affected-area emergency material allocation problem with time-varying supply-demand by topological structure transformation, and an optimized combinatorial algorithm was designed.(5) Single-R2DC and multi-affected-area emergency material allocation problem with time-varying supply-demand constraint and response time constraint was studied. Firstly, the scheme configuration of this problem was addressed. Secondly, based on the analysis of the relationship between system response time and response time constraint of each affected-area, a methodology was suggested by modification on scheme without response time constraint of each affected-area, and a corresponding algorithm was designed.(6) Time-varying supply-demand constrained problem in complex network structure was studied. Firstly, the scheme configuration of this problem with complex network structure was addressed. Secondly, based on the analysis of the relationship among system response time, earliest emergency response start time of R2DC and response time constraint of each affected-area, and the analysis the relationship between system response time and critical response time, a material allocation sequence was suggested, and a corresponding algorithm was designed.By studying the problems mentioned above, the decision-makers in disaster relief operation could be assisted to construct a systemic knowledge of emergency logistic and emergency material allocation, to gain intelligence supporting and decision-making reference for emergency logistics operating effectively to guarantee emergency material supply. Thus, the total emergency management system may realize the relief objectives with the earliest emergency response start time and the minimum number of R2DC involved, decline the impact caused by large-scale paroxysmal events and natural disasters on social life and minimize the total systematic loss on casualty and property.更多还原