【作者】 毛子骏；

【导师】 费奇；

【作者基本信息】 华中科技大学 ， 系统工程， 2010， 博士

【摘要】 相互关联的基础设施网络系统是现代社会赖以生存维系的基石,而近年来以“911”恐怖袭击、卡特里娜飓风为代表的各类人为或自然灾害频繁爆发并在基础设施网络中扩散造成严重影响。以美国总统关键设施保护委员会为代表的众多权威机构提出应大力开展基础设施网络中灾害扩散与控制问题的研究,这既是国家安全的现实需要,也具有重要的实际应用价值。本文从单一基础设施网络的研究入手,以基础设施网络的关联关系为重点,以华中地区某主要城市中供电、供水网络组成的关联基础设施网络为案例,以2008年我国爆发的冰雪灾害为背景,建立关联基础设施网络模型,分析了蓄意攻击灾害与自然灾害在基础设施网络中的扩散情形,从基础设施网络内部负载调节与外部资源调度的角度设计了各种灾害控制策略,并通过仿真的方法对灾害扩散影响与灾害控制效果进行了讨论。主要内容为：(1)以华中地区某主要城市的供水、供电网络为案例,分别从结构和功能两个方面分析了灾害在基础设施网络中的扩散与控制问题。发现该市电网具有较高簇聚系数与较短平均路径,且对蓄意攻击抵抗能力较差对随机攻击抵抗能力较好；该市水网具有较小簇聚系数与较短平均路径,且对随机攻击与蓄意攻击都没有较好的抵抗能力。随后建立了该市供电网络的功能负载模型,针对三种灾害类型设计了三种节点容量的控制策略,通过仿真发现优先攻击大负载节点的灾害扩散影响最为严重,优先增加大负载节点容量的控制策略效果最优。最后提出了一种基于节点负载状况的节点修复策略,并利用仿真验证了该策略的有效性。(2)结合单一基础设施网络的运转特性及网络间关联关系的特点提出了一种关联基础设施网络研究模型。与目前研究中采用的Agent、投入产出、系统动力学等模型相比,该模型能更好的从基础设施网络结构及其运转负载的角度分析灾害扩散对基础设施网络性能的影响。在本文后继研究中,通过应用该模型分析蓄意攻击灾害与自然灾害在关联基础设施网络中的扩散与控制问题验证了该模型的有效性。(3)提出应对关联基础设施网络中蓄意攻击灾害的关键就是找准网络中的重要节点。围绕这个问题,提出了一种基于基础设施网络损失代价函数的重要节点判别方法,发现基础设施网络间关联关系的建立不仅会放大灾害的扩散影响,还会改变基础设施网络内重要节点的排序。基于此,提出了一种基于节点度及网络间关联关系的重要节点搜索方法,通过仿真检验发现该方法能很好地缩小搜索范围。最后从控制节点攻击风险的角度给出了一种节点保护资源的分配策略并分析了该策略的有效性。(4)提出了一套分析自然灾害在关联基础设施网络中扩散与控制问题的方法,以华中地区某主要城市2008年遭遇的冰雪灾害为例,验证了该方法的有效性。设计仿真实验模拟了冰雪灾害的扩散过程,发现灾害在网络间的关联扩散给基础设施网络造成了更大的损失。从基础设施网络损失与服务需求满足的角度,对灾害发生后关联基础设施网络的四种负载调整策略进行了仿真分析,发现以本网络损失代价最小为目标的负载调整策略对灾害的控制效果最差,而优先考虑网络关联关系的负载调整策略效果最优。最后从利用外界救灾资源来控制灾害扩散的角度提出了三种救灾资源的分配策略,并通过仿真分析了各种策略的有效性,发现优先修复重要关联节点的策略效果最优。更多还原

【Abstract】 Interdependent infrastructures are the cornerstone of modern society. In recent years, various types of man-made or natural disasters, such as the "911" terrorist attacks and Hurricane Katrina, frequently break out and spread in infrastructures. Many authoritative agencies, such as President’s Commission on Critical Infrastructure Protection (PCCIP), have suggested that the research of disaster spread and control in infrastructures should be vigorously carried out. This is not only the reality need of national security, but also has important practical application value.The dissertation, beginning with analysis of single infrastructure network, is focused on the interdependence study of infrastructures. By taking the snow disaster which broke out in infrastructures in a major city in Central China in 2008 as a study case, it establishes the interdependent infrastructure network model and analyses the spread of deliberate attack and natural disaster in infrastructures. Several kinds of disaster control strategies are proposed based upon load adjustment and rescue resource allocation, of which the spread impact and control effect are discussed based upon simulations. The main contents are summarized as follows:(1) By taking the interdependent infrastructure network consisting of water supply network and power supply network of a major city in Central China as a study case, the dissertation analyzes the problem of disaster spread and control in infrastructures through both structure and function. It is found that the power supply network has high clustering coefficient and short average path length, and the network exhibits a very high degree of robustness against random attack but a low robustness against deliberate attack. It is also found that the water supply network has low clustering coefficient and short average path length, and the network exhibits a low degree of robustness against random attack and deliberate attack. After that, the function model of power supply network of the city is established and simulated for three kinds of disaster control strategies for three kinds of disasters. The simulation result shows that the disaster diffusion effect is of most severe degree if the attack on the big load node is of high priority, and it is most effective for the disaster control to prioritize increasing the capacity of the node with heavy load. Furthermore, based on the node load, the dissertation proposes a node repair strategy and validates it by simulation.(2) The dissertation proposes a model to study interdependent infrastructure networks according to the operation characteristics of single infrastructure and the interdependence among infrastructures. Compared with related researches, such as Agent, Input-Output, System Dynamics model, this model is able to analyze the spread impact of disaster better according to structure and operation characteristics of infrastructures. In the follow-up study, the effectiveness of the model is validated by applying it to analyze the spread and control problem for the impact of deliberate attack and natural disaster upon interdependent infrastructure network.(3) It is pointed out in this dissertation that it is crucial to locate the key node of the infrastructure network for dealing with the deliberate attack on it. Based on this assumption, a method of determining important nodes in infrastructure network based on infrastructure loss function is proposed. It is found that the connection established among infrastructure networks will both amplify the disaster spread impact, and alter the order of important nodes within any infrastructure network. To this end, the dissertation proposes a method to search the important nodes based on degree of node and interdependency of network. The simulation shows that it can effectively narrow the search scope. Finally, a resource allocation strategy for controlling the node attack risk is proposed, and its effectiveness is also analyzed.(4) The dissertation proposes a method to study the spread and control problems of natural disasters in interdependent infrastructures, and the method is validated by applying it to analyze the snow disaster which broke out in infrastructures of a major city in Central China in 2008.Firstly, by simulating the spread of snow disaster in interdependent infrastructures, it shows that the interdependent spread among infrastructure networks brings greater losses for infrastructure network. Secondly, four kinds of load adjustment strategies are simulated for infrastructure network when disaster breaks out for the sake of infrastructure loss and satisfaction of service demand. The simulation result shows that the strategy which aims at minimizing the loss of single infrastructure is the worst, and the strategy which prioritizes the interdependency of infrastructures is the best. Finally, the dissertation proposes three kinds of rescue resource allocation strategies to prevent disaster from spreading, and analyzes the effectiveness of these strategies through simulations. The simulation result shows that the strategy which prioritizes interdependent nodes yields the best performance.更多还原